From b71491c6c910c15bc1580e9469b6d3ecb3b376b1 Mon Sep 17 00:00:00 2001 From: Siddhant Badola Date: Fri, 30 Oct 2026 09:27:59 -0400 Subject: [PATCH 01/18] fix(bedrock): route a Bedrock model id with no Invoke provider to Converse (#43290) * fix(bedrock): route unmapped converse-only family model ids to converse #42713 stopped Bedrock-native OpenAI ids that are not in the cost map yet from falling to the invoke route. The same gap exists for the other families that have no invoke transformation at all: minimax, nvidia, writer, zai, google and xai. An id from one of them that is newer than the cost map hits get_bedrock_route -> "invoke" -> get_bedrock_invoke_provider() -> None -> BedrockError, when Converse would have served it Add a second family regex for those six and default them to converse, the same way the openai family is handled. _OPENAI_FAMILY_MODEL_RE is left alone because the prompt-caching rule is openai-specific Every shipped id of these families already resolves to a converse entry via its base model (checked against the cost map), so this only changes behaviour for ids that are not in the map yet Tests: six unmapped ids route to converse, and an unmapped meta.* id still routes to invoke since meta has an invoke handler * fix(bedrock): route a model id with no Invoke provider to Converse A bedrock/ id missing from the cost map fell to the Invoke route, where an id that names no Invoke provider (a new vendor's model, a bare provisioned-model, imported-model, or custom-model-deployment ARN) fails locally with "Unknown provider=None" before any request is sent. Such ids now go to Converse, which replaces the per-family regex with the general rule. Ids that name an Invoke provider and explicit route prefixes are unchanged. * test(bedrock): expect Converse for ids with no Invoke provider in the nova ARN and not-found tests * fix(bedrock): accept tools on Converse for a model with no cost-map row or Invoke provider, and keep a 404 for an invalid id * fix(bedrock): answer 404 for an invalid model id instead of AWS's 400 AWS answers a mistyped Bedrock model id with a 400 ValidationException. The mapping already raised NotFoundError, but passing AWS's 400 response through made the openai base class report 400, so the proxy answered 400 and the router skipped its cooldown * fix(bedrock): drop lookaround regexes from tool schemas for a model the cost map does not list * fix(bedrock): apply the unknown-model defaults only to bare ids LiteLLM knows nothing about A bare Bedrock id with no cost-map row and no Invoke vendor gets Converse's open params, its reasoning_effort as sent, lookaround-free tool schemas, and a 404 for AWS's invalid-identifier reply, which keeps AWS's response headers. An explicit route like converse/, an ARN, or an id LiteLLM recognizes keeps main's params, regexes and error status, so drop_params still drops tools there * fix(bedrock): judge lookaround support by model id on every route, ignore price-only rows The regex check now strips route and region prefixes from the model and its base_model before asking whether LiteLLM knows the model, so bedrock/converse/ gets the same answer in a direct transform call as on the completion path. A cost-map row counts only when it names a provider, so a price-only register_model row leaves an unlisted id unknown, as on the proxy. AmazonInvokeConfig now inherits the Invoke provider lookup instead of keeping an identical copy. * fix(bedrock): treat a region path like the bare id when the cost map does not list it A deployment like bedrock/us-east-1/ now gets the same tools, reasoning_effort and typo 404 as bedrock/, so Claude Code works on it and a mistyped one still answers 404. * fix(bedrock): keep today's 404 for ids that reach Converse only by default, and keep an empty id local * fix(bedrock): answer 404 for an invalid id only when Converse rejected it, and keep a region path with no id local * fix(bedrock): keep a path-shaped id local, and map /v1/completions errors with the model's provider An id with a slash that is not an ARN, like bedrock/../../converse or an unknown region path, keeps Invoke and its local 404 instead of reaching AWS Converse. atext_completion mapped every error as OpenAI, so a Bedrock typo on /v1/completions answered 400 while chat answered 404. It now resolves the provider from the request's model, as the other async entry points do. * fix(text completion): keep /v1/completions error mapping as it is on main Mapping async /v1/completions errors with each provider's rules is wider than this Bedrock routing change. It moves classes, statuses, cooldown and fallbacks for most providers, and some lose context-window fallbacks. So this restores main's mapping here and leaves that change to its own PR. A Bedrock typo on /v1/completions answers AWS's 400, while chat keeps the 404. --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> --- .../exception_mapping_utils.py | 32 +- .../bedrock/chat/converse_transformation.py | 8 +- .../base_invoke_transformation.py | 62 +--- litellm/llms/bedrock/common_utils.py | 68 +++- .../test_bedrock_kimi_k3_cache_point_wire.py | 2 +- .../test_exception_mapping_utils.py | 70 +++- .../chat/test_converse_transformation.py | 312 +++++++++++++++++- .../llms/bedrock/test_bedrock_common_utils.py | 107 +++++- .../llms/bedrock/test_nova_imported_models.py | 6 +- 9 files changed, 559 insertions(+), 108 deletions(-) diff --git a/litellm/litellm_core_utils/exception_mapping_utils.py b/litellm/litellm_core_utils/exception_mapping_utils.py index 5a27f5446ad..33b6d4e1857 100644 --- a/litellm/litellm_core_utils/exception_mapping_utils.py +++ b/litellm/litellm_core_utils/exception_mapping_utils.py @@ -18,6 +18,7 @@ from litellm.litellm_core_utils.bug_report import ( ) from litellm.litellm_core_utils.core_helpers import as_str_mapping, attribute_or_none from litellm.litellm_core_utils.secret_redaction import redact_string as redact_secret_string +from litellm.llms.bedrock.common_utils import bedrock_model_reaches_converse_by_default from litellm.llms.bigmodel.common_utils import raise_bigmodel_business_error from litellm.types.utils import LlmProviders @@ -938,13 +939,22 @@ _BEDROCK_MANTLE_CONTEXT_WINDOW_GENERIC_MESSAGE: Final = ( ) -def _bedrock_mantle_context_window_response(original_exception: _ProviderHTTPException) -> httpx.Response: +def _bedrock_response_with_status(original_exception: _ProviderHTTPException, status_code: int) -> httpx.Response: original_response: Final = getattr(original_exception, "response", None) if not isinstance(original_response, httpx.Response) or getattr(original_response, "_request", None) is None: - return httpx.Response(status_code=400, request=httpx.Request(method="POST", url="https://api.openai.com/v1/")) - if original_response.status_code == 400: + return httpx.Response( + status_code=status_code, request=httpx.Request(method="POST", url="https://api.openai.com/v1/") + ) + if original_response.status_code == status_code: return original_response - return httpx.Response(status_code=400, headers=original_response.headers, request=original_response.request) + return httpx.Response(status_code=status_code, headers=original_response.headers, request=original_response.request) + + +def _came_from_bedrock_converse(original_exception: _ProviderHTTPException) -> bool: + original_response: Final = getattr(original_exception, "response", None) + if not isinstance(original_response, httpx.Response) or getattr(original_response, "_request", None) is None: + return False + return original_response.request.url.path.endswith(("/converse", "/converse-stream")) def _get_bedrock_mantle_context_window_message(error_str: str) -> str | None: @@ -980,7 +990,7 @@ def _map_bedrock_exception( message=mantle_context_window_message, model=model, llm_provider=custom_llm_provider, - response=_bedrock_mantle_context_window_response(original_exception), + response=_bedrock_response_with_status(original_exception, 400), ) if ( "too many tokens" in error_str @@ -1003,6 +1013,18 @@ def _map_bedrock_exception( llm_provider="bedrock", response=getattr(original_exception, "response", None), ) + elif ( + "The provided model identifier is invalid" in error_str + and bool(model) + and _came_from_bedrock_converse(original_exception) + and bedrock_model_reaches_converse_by_default(model) + ): + raise NotFoundError( + message=f"BedrockException - {error_str}", + model=model, + llm_provider="bedrock", + response=_bedrock_response_with_status(original_exception, 404), + ) elif "Malformed input request" in error_str: raise BadRequestError( message=f"BedrockException - {error_str}", diff --git a/litellm/llms/bedrock/chat/converse_transformation.py b/litellm/llms/bedrock/chat/converse_transformation.py index 443e722168d..5e621a5f322 100644 --- a/litellm/llms/bedrock/chat/converse_transformation.py +++ b/litellm/llms/bedrock/chat/converse_transformation.py @@ -63,6 +63,7 @@ from litellm.llms.bedrock.chat.tool_result_images import place_tool_result_image from litellm.llms.bedrock.common_utils import ( bedrock_chat_uses_native_structured_output, bedrock_cleaned_output_schema, + bedrock_model_is_unknown, bedrock_model_supports_regex_lookaround, bedrock_native_output_format, bedrock_native_output_is_opt_in, @@ -529,13 +530,14 @@ class AmazonConverseConfig(BaseConfig): """ Handle the reasoning_effort parameter based on the model type. - - GPT-OSS and DeepSeek V3 models: passed through unchanged via additionalModelRequestFields. + - GPT-OSS, DeepSeek V3 and unknown models (``bedrock_model_is_unknown``): passed through unchanged + via additionalModelRequestFields. - OpenAI GPT-5.x and GPT-6 models: mapped to ``reasoning.effort`` via additionalModelRequestFields. - Nova 2 models: transformed to reasoningConfig. - Anthropic models: mapped to ``thinking`` (and ``output_config.effort`` on adaptive Claude 4.6 / 4.7). """ - if "gpt-oss" in model or "deepseek" in model: + if "gpt-oss" in model or "deepseek" in model or bedrock_model_is_unknown(model): optional_params["reasoning_effort"] = reasoning_effort elif self._is_openai_gpt_reasoning_model(model): reasoning: Final[BedrockConverseGptReasoningEffortBlock] = {"effort": reasoning_effort} @@ -681,7 +683,7 @@ class AmazonConverseConfig(BaseConfig): "parallel_tool_calls", ] - if "arn" in model: # we can't infer the model from the arn, so just add all params + if "arn" in model or bedrock_model_is_unknown(model): supported_params.append("tools") supported_params.append("tool_choice") supported_params.append("thinking") diff --git a/litellm/llms/bedrock/chat/invoke_transformations/base_invoke_transformation.py b/litellm/llms/bedrock/chat/invoke_transformations/base_invoke_transformation.py index 852179e8caa..877fb4ca692 100644 --- a/litellm/llms/bedrock/chat/invoke_transformations/base_invoke_transformation.py +++ b/litellm/llms/bedrock/chat/invoke_transformations/base_invoke_transformation.py @@ -2,7 +2,7 @@ import copy import json import time from collections.abc import Mapping, Sequence -from typing import TYPE_CHECKING, Any, Final, cast, get_args +from typing import TYPE_CHECKING, Any, Final import httpx from pydantic import ConfigDict, TypeAdapter, ValidationError, with_config @@ -74,10 +74,6 @@ _GUARDRAIL_CONFIG_EXPECTED_FORMAT: Final = ( "{'guardrailIdentifier': str, 'guardrailVersion': str, 'trace': 'enabled'|'disabled'|'enabled_full'}" ) -_INVOKE_PROVIDERS: Final = TypeAdapter(tuple[litellm.BEDROCK_INVOKE_PROVIDERS_LITERAL, ...]).validate_python( - get_args(litellm.BEDROCK_INVOKE_PROVIDERS_LITERAL) -) - def _response_header(headers: Mapping[str, str], name: str) -> str | None: return headers.get(name, None) @@ -562,62 +558,6 @@ class AmazonInvokeConfig(BaseConfig, BaseAWSLLM): """ return False - @staticmethod - def get_bedrock_invoke_provider( - model: str, - ) -> litellm.BEDROCK_INVOKE_PROVIDERS_LITERAL | None: - """ - Helper function to get the bedrock provider from the model - - handles 4 scenarios: - 1. model=invoke/anthropic.claude-3-5-sonnet-20240620-v1:0 -> Returns `anthropic` - 2. model=anthropic.claude-3-5-sonnet-20240620-v1:0 -> Returns `anthropic` - 3. model=llama/arn:aws:bedrock:us-east-1:086734376398:imported-model/r4c4kewx2s0n -> Returns `llama` - 4. model=us.amazon.nova-pro-v1:0 -> Returns `nova` - """ - if model.startswith("invoke/"): - model = model.replace("invoke/", "", 1) - - # Special case: Check for "nova" in model name first (before "amazon") - # This handles amazon.nova-* models which would otherwise match "amazon" (Titan) - if "nova" in model.lower(): - if "nova" in get_args(litellm.BEDROCK_INVOKE_PROVIDERS_LITERAL): - return cast(litellm.BEDROCK_INVOKE_PROVIDERS_LITERAL, "nova") - - _split_model: Final = model.split(".")[0] - if _split_model in get_args(litellm.BEDROCK_INVOKE_PROVIDERS_LITERAL): - return cast(litellm.BEDROCK_INVOKE_PROVIDERS_LITERAL, _split_model) - - # If not a known provider, check for pattern with two slashes - provider = AmazonInvokeConfig._get_provider_from_model_path(model) - if provider is not None: - return provider - - for provider in _INVOKE_PROVIDERS: - if provider in model: - return provider - return None - - @staticmethod - def _get_provider_from_model_path( - model_path: str, - ) -> litellm.BEDROCK_INVOKE_PROVIDERS_LITERAL | None: - """ - Helper function to get the provider from a model path with format: provider/model-name - - Args: - model_path (str): The model path (e.g., 'llama/arn:aws:bedrock:us-east-1:086734376398:imported-model/r4c4kewx2s0n' or 'anthropic/model-name') - - Returns: - Optional[str]: The provider name, or None if no valid provider found - """ - parts: Final = model_path.split("/") - if len(parts) >= 1: - provider: Final = parts[0] - if provider in get_args(litellm.BEDROCK_INVOKE_PROVIDERS_LITERAL): - return cast(litellm.BEDROCK_INVOKE_PROVIDERS_LITERAL, provider) - return None - def convert_messages_to_prompt(self, model: str, messages, provider, custom_prompt_dict) -> tuple[str, list | None]: # handle anthropic prompts and amazon titan prompts prompt = "" diff --git a/litellm/llms/bedrock/common_utils.py b/litellm/llms/bedrock/common_utils.py index d88d7e6280c..d6f5d5b6bd7 100644 --- a/litellm/llms/bedrock/common_utils.py +++ b/litellm/llms/bedrock/common_utils.py @@ -1098,6 +1098,11 @@ def _model_cost_entry(key: str) -> Mapping[str, object] | None: return None if raw is None else _MODEL_COST_ENTRY_ADAPTER.validate_python(raw) +def _cost_map_row_names_a_provider(key: str) -> bool: + entry: Final = _model_cost_entry(key) + return entry is not None and entry.get("litellm_provider") is not None + + def _bedrock_price_map_entries(model: str) -> tuple[Mapping[str, object] | None, ...]: return tuple( _model_cost_entry(key) @@ -1567,7 +1572,8 @@ def bedrock_model_supports_regex_lookaround(model: str, litellm_params: Mapping[ The deployment's ``model_info.supports_regex_lookaround`` wins, then the ``model_prices_and_context_window.json`` entry of its ``base_model``, then the - entry of ``model`` itself. A model nobody flagged keeps its schema as sent. + entry of ``model`` itself. A model nobody flagged keeps its schema as sent, unless + no candidate names a model LiteLLM knows, whatever route prefix it carries. """ params: Final = litellm_params or {} deployment_flag: Final = deployment_model_info(params).get("supports_regex_lookaround") @@ -1576,7 +1582,41 @@ def bedrock_model_supports_regex_lookaround(model: str, litellm_params: Mapping[ base_model: Final = params.get("base_model") candidates: Final = (*((base_model,) if isinstance(base_model, str) else ()), model) flags: Final = (_bedrock_converse_model_flag(candidate, "supports_regex_lookaround") for candidate in candidates) - return next((flag for flag in flags if flag is not None), True) + flag: Final = next((flag for flag in flags if flag is not None), None) + if flag is not None: + return flag + return not all(bedrock_model_is_unknown(split_bedrock_region_path(candidate)[1]) for candidate in candidates) + + +def _bedrock_model_id_without_region(model: str) -> str: + routed_id: Final = model.removeprefix("bedrock/") + region, separator, regional_id = routed_id.partition("/") + return regional_id if separator and region in get_all_bedrock_regions() else routed_id + + +def bedrock_model_is_unknown(model: str) -> bool: + model_id: Final = _bedrock_model_id_without_region(model) + if "/" in model_id: + return False + base: Final = get_bedrock_base_model(model_id) + regional_ids: Final = (f"{region}.{base}" for region in get_bedrock_cross_region_inference_regions()) + if any(_cost_map_row_names_a_provider(key) for key in (*_bedrock_model_info_candidates(model_id), *regional_ids)): + return False + from litellm.llms.bedrock.base_aws_llm import BaseAWSLLM + + return BaseAWSLLM.get_bedrock_invoke_provider(model_id) is None + + +def _names_no_invoke_provider(model: str) -> bool: + from litellm.llms.bedrock.base_aws_llm import BaseAWSLLM + + model_id: Final = _bedrock_model_id_without_region(model) + names_one_model: Final = bool(model_id) and ("/" not in model_id or model_id.startswith("arn:")) + return names_one_model and BaseAWSLLM.get_bedrock_invoke_provider(model) is None + + +def bedrock_model_reaches_converse_by_default(model: str) -> bool: + return BedrockModelInfo.named_bedrock_route(model, None) is None and _names_no_invoke_provider(model) _BedrockConverseModelFlag: TypeAlias = Literal[ @@ -1585,10 +1625,13 @@ _BedrockConverseModelFlag: TypeAlias = Literal[ ] -def _bedrock_converse_model_flag(model: str, key: _BedrockConverseModelFlag) -> bool | None: +def _bedrock_model_info_candidates(model: str) -> tuple[str, ...]: base: Final = get_bedrock_base_model(model) - candidates: Final = dict.fromkeys((model, base, _BEDROCK_MODEL_VERSION_SUFFIX_RE.sub("", base))) - for candidate in candidates: + return tuple(dict.fromkeys((model, base, _BEDROCK_MODEL_VERSION_SUFFIX_RE.sub("", base)))) + + +def _bedrock_converse_model_flag(model: str, key: _BedrockConverseModelFlag) -> bool | None: + for candidate in _bedrock_model_info_candidates(model): with contextlib.suppress(Exception): model_info = get_cached_model_info()( model=candidate, @@ -1766,8 +1809,19 @@ class BedrockModelInfo(BaseLLMModelInfo): default (``bedrock_runtime_chat_completions_is_default``) and ``chat_completions/`` opts any other model in; ``request_params`` (the caller's chat params) sends such a request to Converse when it needs a feature only Converse serves, and ``converse/`` pins a model to Converse. Every other OpenAI-family - model stays on Converse without the prefix. + model stays on Converse without the prefix. A model the cost map does not list as Converse goes to + Invoke only when its id names an Invoke provider, and to Converse otherwise. """ + named_route: Final = BedrockModelInfo.named_bedrock_route(model, request_params) + if named_route is not None: + return named_route + return "converse" if _names_no_invoke_provider(model) else "invoke" + + @staticmethod + def named_bedrock_route( + model: str, + request_params: Mapping[str, object] | None, + ) -> BedrockRoute | None: route_mappings: dict[ str, Literal[ @@ -1822,7 +1876,7 @@ class BedrockModelInfo(BaseLLMModelInfo): return "converse" if _OPENAI_FAMILY_MODEL_RE.search(base_model): return "converse" - return "invoke" + return None @staticmethod def _explicit_converse_route(model: str) -> bool: diff --git a/tests/integration/providers/test_bedrock_kimi_k3_cache_point_wire.py b/tests/integration/providers/test_bedrock_kimi_k3_cache_point_wire.py index 854da6fce0b..0772e928b35 100644 --- a/tests/integration/providers/test_bedrock_kimi_k3_cache_point_wire.py +++ b/tests/integration/providers/test_bedrock_kimi_k3_cache_point_wire.py @@ -1037,7 +1037,7 @@ def test_s07_a_5kb_caller_model_under_the_wildcard_is_a_bounded_4xx_and_liveline error: Final = _JSON.validate_python(_JSON.validate_python(json.loads(outcome.raw))["error"]) assert "a" * 5120 in str(error["message"]), outcome.raw[:300] assert_within_factor_of_control(oversized.seconds, control=control.seconds, factor=20) - assert rig.wire.drain() == () + assert len(rig.wire.drain()) == 1 _failure_row(outcome.call_id) diff --git a/tests/unit/litellm_core_utils/test_exception_mapping_utils.py b/tests/unit/litellm_core_utils/test_exception_mapping_utils.py index 88501a23235..efb9648cf13 100644 --- a/tests/unit/litellm_core_utils/test_exception_mapping_utils.py +++ b/tests/unit/litellm_core_utils/test_exception_mapping_utils.py @@ -60,7 +60,6 @@ def test_openai_cyber_policy_uses_the_structured_error_code(shape: str, code: st assert caught.value.response.headers["x-request-id"] == "cyber-policy-request" - def _debug_info_for_api_base(configured_api_base: str) -> str: request: Final = httpx.Request("POST", "https://gateway.example/v1/chat/completions") original: Final = OpenAIError( @@ -1602,6 +1601,75 @@ def test_bedrock_mantle_400_maps_to_bad_request(): assert type(excinfo.value) is litellm.BadRequestError +class _ErrorWithoutResponse(Exception): + def __init__(self, message: str, status_code: int) -> None: + super().__init__(message) + self.message = message + self.status_code = status_code + + +def test_bedrock_mantle_context_overflow_without_a_provider_response_answers_400(): + with pytest.raises(litellm.ContextWindowExceededError) as excinfo: + exception_type( + model="openai.gpt-5.6-sol", + original_exception=_ErrorWithoutResponse( + '{"error":{"code":"context_length_exceeded","message":"too long"}}', 400 + ), + custom_llm_provider="bedrock_mantle", + ) + + assert excinfo.value.response.status_code == 400 + + +def _invalid_model_identifier_from(url: str) -> BedrockError: + return BedrockError( + status_code=400, + message='{"message":"The provided model identifier is invalid."}', + response=httpx.Response(status_code=400, request=httpx.Request(method="POST", url=url)), + ) + + +@pytest.mark.parametrize( + "path", + ["/model/us.newvendor.typo-1/converse", "/model/us.newvendor.typo-1/converse-stream"], +) +def test_bedrock_converse_rejecting_an_id_with_no_invoke_provider_answers_404(path): + with pytest.raises(litellm.NotFoundError) as excinfo: + exception_type( + model="us.newvendor.typo-1", + original_exception=_invalid_model_identifier_from(f"https://bedrock-runtime.us-east-1.amazonaws.com{path}"), + custom_llm_provider="bedrock", + ) + + assert excinfo.value.response.status_code == 404 + + +@pytest.mark.parametrize( + "original_exception", + [ + _invalid_model_identifier_from("https://bedrock-runtime.us-east-1.amazonaws.com/model/claude-typo-9/invoke"), + _ErrorWithoutResponse("The provided model identifier is invalid.", 400), + ], + ids=["invoke", "no-response"], +) +def test_bedrock_invalid_model_identifier_not_from_converse_stays_a_bad_request(original_exception): + with pytest.raises(litellm.BadRequestError) as excinfo: + exception_type(model="claude-typo-9", original_exception=original_exception, custom_llm_provider="bedrock") + + assert excinfo.value.status_code == 400 + + +def test_bedrock_invalid_model_identifier_without_a_model_stays_a_bad_request(): + with pytest.raises(litellm.BadRequestError) as excinfo: + exception_type( + model=None, + original_exception=_ErrorWithoutResponse("The provided model identifier is invalid.", 400), + custom_llm_provider="bedrock", + ) + + assert excinfo.value.status_code == 400 + + def test_bedrock_mantle_context_overflow_maps_to_context_window_exceeded(): from litellm.llms.base_llm.chat.transformation import BaseLLMException diff --git a/tests/unit/llms/bedrock/chat/test_converse_transformation.py b/tests/unit/llms/bedrock/chat/test_converse_transformation.py index d82ba192b1f..743e7e74431 100644 --- a/tests/unit/llms/bedrock/chat/test_converse_transformation.py +++ b/tests/unit/llms/bedrock/chat/test_converse_transformation.py @@ -1221,6 +1221,97 @@ def test_transform_request_drops_lookaround_regex_when_the_deployment_model_info assert properties == _LOOKAROUND_FREE_PROPERTIES +@pytest.mark.parametrize( + ("model", "litellm_params"), + [ + ("us.newvendor.model-1", None), + ("converse/us.newvendor.model-1", None), + ("us-east-1/us.newvendor.model-1", None), + ("us.newvendor.model-1", {"base_model": "bedrock/converse/us.newvendor.model-1"}), + ], +) +def test_transform_request_drops_lookaround_regex_for_a_model_the_cost_map_does_not_list(model, litellm_params): + assert "us.newvendor.model-1" not in litellm.model_cost + + properties: Final = _tool_schema_properties(model, _ARTIFACT_DATA_ANTHROPIC_TOOL, litellm_params) + + assert properties == _LOOKAROUND_FREE_PROPERTIES + + +@pytest.mark.parametrize( + "model", + [ + "arn:aws:bedrock:us-east-1:123456789012:custom-model-deployment/abc123", + "arn:aws:bedrock:us-east-1:123456789012:application-inference-profile/abc123", + "us.anthropic.claude-unreleased-v1:0", + "eu.openai.gpt-unreleased", + "apac.newvendor.listed-1", + ], +) +def test_transform_request_keeps_lookaround_regex_for_an_unlisted_model_litellm_still_recognizes(model, monkeypatch): + monkeypatch.setitem(litellm.model_cost, "us.newvendor.listed-1", {"litellm_provider": "bedrock_converse"}) + + properties = _tool_schema_properties(model, _ARTIFACT_DATA_ANTHROPIC_TOOL) + + assert properties == _ARTIFACT_DATA_INPUT_SCHEMA["properties"] + + +def test_transform_request_keeps_lookaround_regex_for_an_unlisted_model_whose_deployment_opts_out(): + properties = _tool_schema_properties( + "us.newvendor.model-1", _ARTIFACT_DATA_ANTHROPIC_TOOL, {"model_info": {"supports_regex_lookaround": True}} + ) + + assert properties == _ARTIFACT_DATA_INPUT_SCHEMA["properties"] + + +@pytest.mark.parametrize( + "deployment_params", + [ + {"model": "bedrock/us.newvendor.model-1"}, + {"model": "bedrock/converse/us.newvendor.model-1", "allowed_openai_params": ["tools"]}, + ], +) +def test_a_router_deployment_of_a_model_the_cost_map_does_not_list_drops_lookaround_regex( + respx_mock, deployment_params +): + route: Final = respx_mock.post( + "https://bedrock-runtime.us-east-1.amazonaws.com/model/us.newvendor.model-1/converse" + ).mock( + return_value=httpx.Response( + 200, + json={ + "output": {"message": {"role": "assistant", "content": [{"text": "ok"}]}}, + "stopReason": "end_turn", + "usage": {"inputTokens": 1, "outputTokens": 1, "totalTokens": 2}, + }, + ) + ) + router: Final = Router( + model_list=[ + { + "model_name": "unlisted", + "litellm_params": { + **deployment_params, + "aws_access_key_id": "fake-access-key", + "aws_secret_access_key": "fake-secret-key", + "aws_region_name": "us-east-1", + }, + "model_info": {"input_cost_per_token": 1e-6, "output_cost_per_token": 2e-6}, + } + ] + ) + + router.completion( + model="unlisted", + messages=[{"role": "user", "content": "hi"}], + tools=[_ARTIFACT_DATA_OPENAI_TOOL], + client=HTTPHandler(), + ) + + tool: Final = json.loads(route.calls[0].request.content)["toolConfig"]["tools"][0] + assert tool["toolSpec"]["inputSchema"]["json"]["properties"] == _LOOKAROUND_FREE_PROPERTIES + + def test_transform_request_keeps_lookaround_regex_when_the_deployment_model_info_opts_out(): properties = _tool_schema_properties( "global.moonshotai.kimi-k3", _ARTIFACT_DATA_ANTHROPIC_TOOL, {"model_info": {"supports_regex_lookaround": True}} @@ -9189,18 +9280,223 @@ def test_bedrock_tools_transformation_valid_params(): assert "test" in result[0]["toolSpec"]["inputSchema"]["json"]["required"] -def test_not_found_error(): +def _invalid_model_identifier_reply(sent: list[httpx.Request]): + def respond(request: httpx.Request) -> httpx.Response: + sent.append(request) + return httpx.Response( + 400, + json={"message": "The provided model identifier is invalid."}, + headers={"x-amzn-ErrorType": "ValidationException", "x-amzn-RequestId": "req-123"}, + ) + + return respond + + +@pytest.mark.parametrize( + "model, path", + [ + ("bedrock/bad_model", b"/model/bad_model/converse"), + ("bedrock/us-east-1/bad_model", b"/model/bad_model/converse"), + ("bedrock/newvendor.listed-v1:0", b"/model/newvendor.listed-v1%3A0/converse"), + ("bedrock/us-east-1/newvendor.listed-v1:0", b"/model/newvendor.listed-v1%3A0/converse"), + ], +) +def test_not_found_error(model, path, monkeypatch): + monkeypatch.setitem(litellm.model_cost, "newvendor.listed-v1:0", {"litellm_provider": "bedrock", "mode": "chat"}) + sent: Final[list[httpx.Request]] = [] + + with pytest.raises(litellm.NotFoundError) as exc_info: + completion( + model=model, + messages=[{"role": "user", "content": "What is the meaning of life"}], + client=HTTPHandler( + client=httpx.Client(transport=httpx.MockTransport(_invalid_model_identifier_reply(sent))) + ), + aws_access_key_id="testing", + aws_secret_access_key="testing", + aws_region_name="us-west-2", + ) + + assert exc_info.value.status_code == 404 + assert exc_info.value.response.status_code == 404 + assert exc_info.value.response.headers["x-amzn-requestid"] == "req-123" + assert [request.url.raw_path for request in sent] == [path] + + +@pytest.mark.parametrize( + "model", ["bedrock/", "bedrock/us-east-1/", "bedrock/../../converse", "bedrock/xx-fake-1/us.xai.grok-4.7"] +) +def test_a_model_id_that_is_empty_or_a_path_fails_locally_without_calling_aws(model): + sent: Final[list[httpx.Request]] = [] + with pytest.raises(litellm.NotFoundError): completion( - model="bedrock/bad_model", - messages=[ - { - "role": "user", - "content": "What is the meaning of life", - } - ], + model=model, + messages=[{"role": "user", "content": "What is the meaning of life"}], + client=HTTPHandler( + client=httpx.Client(transport=httpx.MockTransport(_invalid_model_identifier_reply(sent))) + ), + aws_access_key_id="testing", + aws_secret_access_key="testing", + aws_region_name="us-west-2", ) + assert sent == [] + + +@pytest.mark.parametrize( + "model, path", + [ + ("bedrock/converse/bad_model", b"/model/bad_model/converse"), + ("bedrock/anthropic.claude-typo-v1:0", b"/model/anthropic.claude-typo-v1:0/invoke"), + ], +) +def test_invalid_model_identifier_stays_a_bad_request_for_a_model_litellm_recognizes(model, path): + sent: Final[list[httpx.Request]] = [] + + with pytest.raises(litellm.BadRequestError) as exc_info: + completion( + model=model, + messages=[{"role": "user", "content": "What is the meaning of life"}], + client=HTTPHandler( + client=httpx.Client(transport=httpx.MockTransport(_invalid_model_identifier_reply(sent))) + ), + aws_access_key_id="testing", + aws_secret_access_key="testing", + aws_region_name="us-west-2", + ) + + assert exc_info.value.status_code == 400 + assert [request.url.raw_path for request in sent] == [path] + + +_WEATHER_TOOL: Final = { + "type": "function", + "function": { + "name": "get_weather", + "description": "Get the weather for a city", + "parameters": {"type": "object", "properties": {"city": {"type": "string"}}, "required": ["city"]}, + }, +} + + +def _converse_tool_call_reply(sent: list[httpx.Request]): + def respond(request: httpx.Request) -> httpx.Response: + sent.append(request) + return httpx.Response( + 200, + json={ + "output": { + "message": { + "role": "assistant", + "content": [ + {"toolUse": {"toolUseId": "t1", "name": "get_weather", "input": {"city": "Paris"}}} + ], + } + }, + "stopReason": "tool_use", + "usage": {"inputTokens": 1, "outputTokens": 1, "totalTokens": 2}, + }, + ) + + return respond + + +@pytest.mark.parametrize("model", ["bedrock/us.newvendor.model-1", "bedrock/us-east-1/us.newvendor.model-1"]) +def test_model_missing_from_cost_map_sends_tools_to_converse(model): + assert "us.newvendor.model-1" not in litellm.model_cost + sent: Final[list[httpx.Request]] = [] + + response: Final = completion( + model=model, + messages=[{"role": "user", "content": "What is the weather in Paris?"}], + tools=[_WEATHER_TOOL], + tool_choice="auto", + client=HTTPHandler(client=httpx.Client(transport=httpx.MockTransport(_converse_tool_call_reply(sent)))), + aws_access_key_id="testing", + aws_secret_access_key="testing", + aws_region_name="us-west-2", + ) + + assert response.choices[0].message.tool_calls[0].function.name == "get_weather" + assert [request.url.raw_path for request in sent] == [b"/model/us.newvendor.model-1/converse"] + tool_config: Final = json.loads(sent[0].content)["toolConfig"] + assert [tool["toolSpec"]["name"] for tool in tool_config["tools"]] == ["get_weather"] + assert tool_config["toolChoice"] == {"auto": {}} + + +def test_a_price_only_row_for_an_unlisted_model_still_sends_tools_to_converse(monkeypatch): + monkeypatch.setitem( + litellm.model_cost, "us.newvendor.priced-1", {"input_cost_per_token": 1e-6, "output_cost_per_token": 2e-6} + ) + sent: Final[list[httpx.Request]] = [] + + completion( + model="bedrock/us.newvendor.priced-1", + messages=[{"role": "user", "content": "What is the weather in Paris?"}], + tools=[_WEATHER_TOOL], + client=HTTPHandler(client=httpx.Client(transport=httpx.MockTransport(_converse_tool_call_reply(sent)))), + aws_access_key_id="testing", + aws_secret_access_key="testing", + aws_region_name="us-west-2", + ) + + tool_config: Final = json.loads(sent[0].content)["toolConfig"] + assert [tool["toolSpec"]["name"] for tool in tool_config["tools"]] == ["get_weather"] + + +def test_explicit_converse_route_to_an_unlisted_model_still_drops_tools_under_drop_params(): + sent: Final[list[httpx.Request]] = [] + + completion( + model="bedrock/converse/us.newvendor.model-1", + messages=[{"role": "user", "content": "What is the weather in Paris?"}], + tools=[_WEATHER_TOOL], + drop_params=True, + client=HTTPHandler(client=httpx.Client(transport=httpx.MockTransport(_converse_tool_call_reply(sent)))), + aws_access_key_id="testing", + aws_secret_access_key="testing", + aws_region_name="us-west-2", + ) + + assert "toolConfig" not in json.loads(sent[0].content) + + +def test_unknown_model_gets_reasoning_effort_as_sent(): + sent: Final[list[httpx.Request]] = [] + + completion( + model="bedrock/us.newvendor.model-1", + messages=[{"role": "user", "content": "What is the weather in Paris?"}], + reasoning_effort="high", + client=HTTPHandler(client=httpx.Client(transport=httpx.MockTransport(_converse_tool_call_reply(sent)))), + aws_access_key_id="testing", + aws_secret_access_key="testing", + aws_region_name="us-west-2", + ) + + assert json.loads(sent[0].content)["additionalModelRequestFields"] == {"reasoning_effort": "high"} + + +def test_mapped_model_without_function_calling_still_rejects_tools(): + model: Final = "bedrock/converse/meta.llama3-8b-instruct-v1:0" + assert "meta.llama3-8b-instruct-v1:0" in litellm.model_cost + assert not litellm.supports_function_calling(model=model) + sent: Final[list[httpx.Request]] = [] + + with pytest.raises(litellm.UnsupportedParamsError): + completion( + model=model, + messages=[{"role": "user", "content": "What is the weather in Paris?"}], + tools=[_WEATHER_TOOL], + client=HTTPHandler(client=httpx.Client(transport=httpx.MockTransport(_converse_tool_call_reply(sent)))), + aws_access_key_id="testing", + aws_secret_access_key="testing", + aws_region_name="us-west-2", + ) + + assert sent == [] + @pytest.mark.parametrize( "model, expected_base_model", diff --git a/tests/unit/llms/bedrock/test_bedrock_common_utils.py b/tests/unit/llms/bedrock/test_bedrock_common_utils.py index dd89d9c46b6..20c0e10cb81 100644 --- a/tests/unit/llms/bedrock/test_bedrock_common_utils.py +++ b/tests/unit/llms/bedrock/test_bedrock_common_utils.py @@ -214,24 +214,17 @@ def test_system_defined_inference_profile_arn_still_routes_to_converse(): assert route == "converse" -def test_other_opaque_arn_types_still_route_to_invoke(): - """ - Only application-inference-profile ARNs are auto-routed to converse. Other - opaque ARNs (provisioned-model, imported-model, custom-model-deployment) - also yield no invoke provider, but they are frequently invoke-only with - provider-specific body formats, so routing them to converse could break - them. Guards the deliberate scope against an over-broad "any opaque ARN -> - converse" generalization. - """ - for arn_segment in ( - "provisioned-model/abcdefgh1234", - "imported-model/abcdefgh1234", - "custom-model-deployment/abcdefgh1234", - ): - route = BedrockModelInfo.get_bedrock_route( - model=f"bedrock/arn:aws:bedrock:us-east-1:123412341234:{arn_segment}" - ) - assert route == "invoke", f"{arn_segment} should stay on invoke route" +@pytest.mark.parametrize( + "arn_segment", + ["provisioned-model/abcdefgh1234", "imported-model/abcdefgh1234", "custom-model-deployment/abcdefgh1234"], +) +def test_bare_opaque_arn_with_no_invoke_provider_routes_to_converse(arn_segment: str): + from litellm.llms.bedrock.base_aws_llm import BaseAWSLLM + + model: Final = f"bedrock/arn:aws:bedrock:us-east-1:123412341234:{arn_segment}" + assert BaseAWSLLM.get_bedrock_invoke_provider(model) is None + assert BedrockModelInfo.get_bedrock_route(model) == "converse" + assert BedrockModelInfo.get_bedrock_route(f"bedrock/llama/arn:aws:bedrock:us-east-1:123412341234:{arn_segment}") == "invoke" def test_govcloud_cross_region_inference_prefix(): @@ -1098,6 +1091,84 @@ def test_unmapped_openai_family_model_routes_to_converse(): assert BedrockModelInfo.get_bedrock_route(imported) == "openai" +@pytest.mark.parametrize( + "unmapped", + [ + "bedrock/minimax.minimax-m99-unmapped", + "bedrock/us.nvidia.nemotron-99b-unmapped", + "bedrock/writer.palmyra-x99-unmapped", + "bedrock/zai.glm-99-unmapped", + "bedrock/google.gemma-99-unmapped", + "bedrock/global.xai.grok-99-unmapped", + ], +) +def test_unmapped_converse_only_family_model_routes_to_converse(unmapped: str): + import litellm + + assert BedrockModelInfo.get_base_model(unmapped) not in litellm.bedrock_converse_models + assert BedrockModelInfo.get_bedrock_route(unmapped) == "converse" + + +def test_unmapped_model_of_family_with_invoke_handler_still_routes_to_invoke(): + from typing import Final + + import litellm + + unmapped: Final = "bedrock/meta.llama99-unmapped-v1:0" + assert BedrockModelInfo.get_base_model(unmapped) not in litellm.bedrock_converse_models + assert BedrockModelInfo.get_bedrock_route(unmapped) == "invoke" + + +@pytest.mark.parametrize( + ("model", "model_path", "response_json"), + [ + ( + "bedrock/us.newvendor.model-1", + b"/model/us.newvendor.model-1/converse", + { + "output": {"message": {"role": "assistant", "content": [{"text": "ok"}]}}, + "stopReason": "end_turn", + "usage": {"inputTokens": 1, "outputTokens": 1, "totalTokens": 2}, + }, + ), + ( + "bedrock/mistral.mistral-99-unmapped-v1:0", + b"/model/mistral.mistral-99-unmapped-v1:0/invoke", + {"outputs": [{"text": "ok", "stop_reason": "stop"}]}, + ), + ], +) +def test_completion_on_unmapped_model_posts_converse_unless_its_id_names_an_invoke_provider( + model: str, model_path: bytes, response_json: dict[str, object] +): + import httpx + + from litellm.llms.custom_httpx.http_handler import HTTPHandler + + assert BedrockModelInfo.get_base_model(model) not in litellm.bedrock_converse_models + sent: Final[list[httpx.Request]] = [] + + def respond(request: httpx.Request) -> httpx.Response: + sent.append(request) + return httpx.Response( + 200, + json=response_json, + headers={"x-amzn-bedrock-input-token-count": "1", "x-amzn-bedrock-output-token-count": "1"}, + ) + + response: Final = litellm.completion( + model=model, + messages=[{"role": "user", "content": "hello"}], + client=HTTPHandler(client=httpx.Client(transport=httpx.MockTransport(respond))), + aws_access_key_id="testing", + aws_secret_access_key="testing", + aws_region_name="us-west-2", + ) + + assert response.choices[0].message.content == "ok" + assert [request.url.raw_path for request in sent] == [model_path] + + @pytest.mark.parametrize( ("model", "expected"), [ diff --git a/tests/unit/llms/bedrock/test_nova_imported_models.py b/tests/unit/llms/bedrock/test_nova_imported_models.py index e3677aaf9e6..231f3097dcd 100644 --- a/tests/unit/llms/bedrock/test_nova_imported_models.py +++ b/tests/unit/llms/bedrock/test_nova_imported_models.py @@ -24,10 +24,8 @@ class TestGetBedrockRoute: def test_nova2_prefix_routes_to_converse(self): assert BedrockModelInfo.get_bedrock_route(NOVA2_MODEL) == "converse" - def test_plain_arn_routes_to_invoke(self): - # Without spec prefix, ARN doesn't match converse models - result = BedrockModelInfo.get_bedrock_route(f"bedrock/{NOVA_ARN}") - assert result == "invoke" + def test_plain_arn_without_an_invoke_provider_routes_to_converse(self): + assert BedrockModelInfo.get_bedrock_route(f"bedrock/{NOVA_ARN}") == "converse" class TestGetBedrockBaseModel: From cbfb1abdb3fe86a139431360ea7d561579dbfb7f Mon Sep 17 00:00:00 2001 From: Mateo Wang <277851410+mateo-berri@users.noreply.github.com> Date: Fri, 30 Oct 2026 06:58:05 -0700 Subject: [PATCH 02/18] fix(proxy)!: label shared-handler errors the way the route catch-alls do (#39555) * fix(proxy): name shared-handler errors by the status they answer handle_exception_on_proxy pinned type=internal_server_error on every exception it wrapped, so a management route rejecting a caller's own argument answered {"type":"internal_server_error","code":"400"} and an SDK branching on the type read its own bad request as a gateway outage. Derive the type from the status the response actually carries, keeping a type the exception already names. * test(proxy): pin a 500 for exceptions whose status is None * fix(proxy): keep internal_server_error for timeouts and 5xx, redact hosts in shared handler errors Provider 5xx and timeouts keep internal_server_error so retry loops still retry them, the shared handler redacts internal hosts from raw exception messages, and /key/list routes its HTTPException through the shared handler instead of its own copy of the old label. * fix(proxy): label shared-handler errors like the chat route, pick the status before the type, redact only gateway failures * fix(proxy): pass shared-handler error messages through as on main Redacting timeout and 5xx messages changed what every shared-handler route returns, which relabeling the error type does not need, so messages go back to the verbatim text main sends. * fix(proxy): redact only the status-less failure main never sent * fix(proxy): answer a generic 500 for a status-less failure and expect auth_error on the /user/new 403 A failure with no status now answers "Internal server error" instead of its own text, since that text can name a guardrail's private host and pattern redaction misses cluster DNS names and IPv6 addresses. The /user/new integration test now expects the auth_error type main's route catch-alls give a 403 the proxy raises * fix(proxy): answer only a 4xx or 5xx carried status and fall back to the route default otherwise A numeric-string status outside 400 to 599 made /v1/messages, images, rerank and the OpenAI pass-through answer HTTP 200 with an error body or drop the connection, and non-ASCII digits or a 5000-digit string slipped through isdecimal. error_status_code now accepts an int or a three-digit ASCII string in 400 to 599, the range the chat route already enforces * refactor(proxy): parse a three-digit string status without recursing --- .../common_utils/openai_error_payload.py | 48 +++-- .../proxy/guardrails/guardrail_endpoints.py | 3 +- litellm/proxy/utils.py | 35 ++-- .../test_guardrail_submissions.py | 4 +- .../test_user_self_update_guard.py | 2 +- .../test_user_delete_revokes_keys.py | 4 +- .../test_apply_guardrail_endpoint.py | 2 +- .../common_utils/test_openai_error_payload.py | 31 ++- .../credential_endpoints/test_endpoints.py | 17 ++ .../guardrails/test_guardrail_endpoints.py | 35 +++- .../test_key_management_endpoints.py | 2 +- .../proxy/utils/helpers/test_error_helpers.py | 181 +++++++++++++++++- 12 files changed, 311 insertions(+), 53 deletions(-) diff --git a/litellm/proxy/common_utils/openai_error_payload.py b/litellm/proxy/common_utils/openai_error_payload.py index 53b2a05d40f..a948c390375 100644 --- a/litellm/proxy/common_utils/openai_error_payload.py +++ b/litellm/proxy/common_utils/openai_error_payload.py @@ -28,6 +28,8 @@ _OPENAI_ERROR_TYPE_BY_STATUS: Final[Mapping[int, str]] = MappingProxyType( _AUTH_FAILURE_STATUS_CODES: Final = frozenset({status.HTTP_401_UNAUTHORIZED, status.HTTP_403_FORBIDDEN}) +_HTTP_ERROR_STATUSES: Final = range(status.HTTP_400_BAD_REQUEST, 600) + _ERROR_BODY: Final = TypeAdapter(dict[str, object]) _FAILOVER_SQLSTATES: Final = frozenset({"25006", "57P01", "57P02", "57P03"}) @@ -47,13 +49,26 @@ def attribute_of(value: object, name: str, default: object = None) -> object: def error_status_code(exc: object, default: int) -> int: - """The HTTP status an exception carries as ``status_code`` or, the way ``ProxyException`` - stores it, as a stringified ``code``; ``default`` when it carries neither.""" - carried: Final = attribute_of(exc, "status_code") - if isinstance(carried, int) and not isinstance(carried, bool): + """The 4xx or 5xx HTTP status an exception carries as ``status_code`` (an int or a decimal string) or, + the way ``ProxyException`` stores it, as a stringified ``code``; ``default`` when it carries neither.""" + carried: Final = _http_error_status(attribute_of(exc, "status_code")) + if carried is not None: return carried stringified: Final = attribute_of(exc, "code") - return int(stringified) if isinstance(stringified, str) and stringified.isdecimal() else default + from_code: Final = _http_error_status(stringified) if isinstance(stringified, str) else None + return default if from_code is None else from_code + + +def _http_error_status(value: object) -> int | None: + match value: + case bool(): + return None + case int() if value in _HTTP_ERROR_STATUSES: + return value + case str() if len(value) == 3 and value.isascii() and value.isdigit() and int(value) in _HTTP_ERROR_STATUSES: + return int(value) + case _: + return None def openai_error_type(exc: object, status_code: int) -> str: @@ -105,9 +120,9 @@ def proxy_exception_for(exc: Exception, default_status_code: int) -> ProxyExcept code=status.HTTP_503_SERVICE_UNAVAILABLE, ) case HTTPException(status_code=status_code, detail=detail): - return _openai_shaped_proxy_exception(exc, str(detail), status_code) + return openai_shaped_proxy_exception(exc, str(detail), status_code) case _: - return _openai_shaped_proxy_exception( + return openai_shaped_proxy_exception( exc, str(exc) or type(exc).__name__, error_status_code(exc, default_status_code) ) @@ -176,14 +191,19 @@ def _route_failure_error_type(exc: Exception, raised_status: int, answered_statu return openai_error_type(exc, answered_status) -def _openai_shaped_proxy_exception(exc: Exception, message: str, status_code: int) -> ProxyException: +def openai_shaped_proxy_exception(exc: Exception, message: str, status_code: int) -> ProxyException: answered_status: Final = answered_status_code(message, status_code) - return ProxyException( - message=message, - type=_catch_all_error_type(exc, answered_status), - param=openai_error_param(exc), - code=answered_status, - ) + return _proxy_exception(exc, message, answered_status, _catch_all_error_type(exc, answered_status)) + + +def status_typed_proxy_exception(exc: HTTPException) -> ProxyException: + message: Final = str(exc.detail) + answered_status: Final = answered_status_code(message, exc.status_code) + return _proxy_exception(exc, message, answered_status, openai_error_type(exc, answered_status)) + + +def _proxy_exception(exc: Exception, message: str, answered_status: int, error_type: str) -> ProxyException: + return ProxyException(message=message, type=error_type, param=openai_error_param(exc), code=answered_status) def _catch_all_error_type(exc: Exception, status_code: int) -> str: diff --git a/litellm/proxy/guardrails/guardrail_endpoints.py b/litellm/proxy/guardrails/guardrail_endpoints.py index a5822bfb0e3..363b9162e00 100644 --- a/litellm/proxy/guardrails/guardrail_endpoints.py +++ b/litellm/proxy/guardrails/guardrail_endpoints.py @@ -22,6 +22,7 @@ from litellm.litellm_core_utils.safe_json_dumps import safe_dumps from litellm.proxy._types import LitellmUserRoles, UserAPIKeyAuth from litellm.proxy.auth.user_api_key_auth import user_api_key_auth from litellm.proxy.common_utils.callback_utils import CALLBACK_VAR_ENCRYPTED_PREFIX +from litellm.proxy.common_utils.openai_error_payload import status_typed_proxy_exception from litellm.proxy.common_utils.path_utils import is_within, safe_join from litellm.proxy.guardrails.content_filter_data import CATEGORIES_DIR, DATA_ROOTS, category_dirs, find_category_file from litellm.proxy.guardrails.guardrail_hooks.custom_code.bounded_execution import ( @@ -2533,7 +2534,7 @@ async def apply_guardrail( e = transformed_exception except Exception: verbose_proxy_logger.exception("apply_guardrail: post_call_failure_hook failed") - raise handle_exception_on_proxy(e) + raise handle_exception_on_proxy(status_typed_proxy_exception(e) if isinstance(e, HTTPException) else e) # Success logging outside except so a hook error never triggers failure handlers. response = await _emit_guardrail_success_logs( diff --git a/litellm/proxy/utils.py b/litellm/proxy/utils.py index b1055bb9124..74e16628bc8 100644 --- a/litellm/proxy/utils.py +++ b/litellm/proxy/utils.py @@ -85,8 +85,8 @@ from litellm.proxy._types import ( from litellm.proxy.bug_report_config import build_proxy_bug_report from litellm.proxy.common_utils.openai_error_payload import ( attribute_of, - litellm_call_id_headers, - openai_error_param, + error_status_code, + openai_shaped_proxy_exception, with_litellm_call_id, ) from litellm.proxy.spend_tracking.spend_log_error_logger import spend_log_error @@ -9186,39 +9186,30 @@ def _recreate_writer_on_read_only_transaction(prisma_client: "PrismaClient | Non asyncio.create_task(prisma_client.recreate_read_only_writer(reason="postgres_read_only_transaction")) +def _client_message(e: Exception) -> str: + status_is_none: Final = hasattr(e, "status_code") and attribute_of(e, "status_code") is None + return "Internal server error" if status_is_none else strip_bug_report_notice(str(e)) + + def handle_exception_on_proxy(e: Exception, litellm_call_id: str | None = None) -> ProxyException: """ Returns an Exception as ProxyException, this ensures all exceptions are OpenAI API compatible """ - from fastapi import status - verbose_proxy_logger.exception("Exception: %s", e) if PrismaDBExceptionHandler.is_read_only_transaction_error(e): from litellm.proxy.proxy_server import prisma_client _recreate_writer_on_read_only_transaction(prisma_client) - headers: Final = litellm_call_id_headers(litellm_call_id) - if isinstance(e, HTTPException): - return ProxyException( - message=getattr(e, "detail", f"error({e})"), - type=ProxyErrorTypes.internal_server_error, - param=openai_error_param(e), - headers=headers, - code=getattr(e, "status_code", status.HTTP_500_INTERNAL_SERVER_ERROR), - ) - elif isinstance(e, ProxyException): + if isinstance(e, ProxyException): return with_litellm_call_id(e, litellm_call_id) - _status_code: Final = getattr(e, "status_code", status.HTTP_500_INTERNAL_SERVER_ERROR) + if isinstance(e, HTTPException): + return with_litellm_call_id(openai_shaped_proxy_exception(e, str(e.detail), e.status_code), litellm_call_id) if should_report_bug(e): verbose_proxy_logger.error(bug_report_notice(build_proxy_bug_report(e))) - return ProxyException( - message=strip_bug_report_notice(str(e)), - type=ProxyErrorTypes.internal_server_error, - param=openai_error_param(e), - headers=headers, - code=_status_code, - ) + message: Final = _client_message(e) + status_code: Final = error_status_code(e, status.HTTP_500_INTERNAL_SERVER_ERROR) + return with_litellm_call_id(openai_shaped_proxy_exception(e, message, status_code), litellm_call_id) def require_enterprise_license(feature: str | None = None) -> None: diff --git a/tests/integration/authorization/test_guardrail_submissions.py b/tests/integration/authorization/test_guardrail_submissions.py index 16b9cbf31dc..fb095010eb2 100644 --- a/tests/integration/authorization/test_guardrail_submissions.py +++ b/tests/integration/authorization/test_guardrail_submissions.py @@ -432,7 +432,7 @@ def test_team_guardrail_submissions_require_admin_approval( f"Guardrail '{first_name}' not found. " "Please ensure the guardrail is configured in your LiteLLM proxy." ), - "type": "internal_server_error", + "type": "invalid_request_error", "param": None, "code": "404", } @@ -481,7 +481,7 @@ def test_team_guardrail_submissions_require_admin_approval( f"Guardrail '{second_name}' not found. " "Please ensure the guardrail is configured in your LiteLLM proxy." ), - "type": "internal_server_error", + "type": "invalid_request_error", "param": None, "code": "404", } diff --git a/tests/integration/authorization/test_user_self_update_guard.py b/tests/integration/authorization/test_user_self_update_guard.py index 2c495beb77e..4294e8a088d 100644 --- a/tests/integration/authorization/test_user_self_update_guard.py +++ b/tests/integration/authorization/test_user_self_update_guard.py @@ -279,7 +279,7 @@ def test_new_user_with_admin_role_requires_proxy_admin(gateway: Gateway, role: s assert refused.json() == { "error": { "message": f"Only proxy admins can create administrative users (proxy_admin, proxy_admin_viewer). Attempted to create user with role: {role}. Your role: internal_user", - "type": "internal_server_error", + "type": "auth_error", "param": None, "code": "403", } diff --git a/tests/integration/management/test_user_delete_revokes_keys.py b/tests/integration/management/test_user_delete_revokes_keys.py index 513a80b7319..2a1bd210b49 100644 --- a/tests/integration/management/test_user_delete_revokes_keys.py +++ b/tests/integration/management/test_user_delete_revokes_keys.py @@ -170,7 +170,7 @@ def test_user_delete_revokes_keys_and_memberships(gateway: Gateway, peer: Gatewa assert info.json() == { "error": { "message": f"User {user_id} not found", - "type": "internal_server_error", + "type": "invalid_request_error", "param": None, "code": "404", } @@ -337,7 +337,7 @@ def test_scim_user_delete_revokes_keys(gateway: Gateway, peer: Gateway, monkeypa expected_info: Final = { "error": { "message": f"User {user_id} not found", - "type": "internal_server_error", + "type": "invalid_request_error", "param": None, "code": "404", } diff --git a/tests/integration/observability/test_apply_guardrail_endpoint.py b/tests/integration/observability/test_apply_guardrail_endpoint.py index 9fa7d1a74db..f50679bd54e 100644 --- a/tests/integration/observability/test_apply_guardrail_endpoint.py +++ b/tests/integration/observability/test_apply_guardrail_endpoint.py @@ -156,7 +156,7 @@ def test_apply_guardrail_returns_a_block_error_for_generic_guardrails(gateway: G assert error.model_dump() == { "error": { "message": _BLOCK_REASON, - "type": "internal_server_error", + "type": "invalid_request_error", "param": None, "code": 400, } diff --git a/tests/unit/proxy/common_utils/test_openai_error_payload.py b/tests/unit/proxy/common_utils/test_openai_error_payload.py index acefc4c1e2c..d4ab811d505 100644 --- a/tests/unit/proxy/common_utils/test_openai_error_payload.py +++ b/tests/unit/proxy/common_utils/test_openai_error_payload.py @@ -111,10 +111,13 @@ def test_the_default_status_stands_when_the_exception_carries_none(default: int) assert error_status_code(ValueError("boom"), default) == default -@pytest.mark.parametrize("carried_status", [True, False, "429", None, 429.0]) -def test_a_non_int_carried_status_falls_back_to_the_default(carried_status: object): - """True is an int in Python but not an HTTP status, and a stringified one would break - every caller that compares the code numerically.""" +@pytest.mark.parametrize( + "carried_status", + [True, False, "slow down", "", None, 429.0, 200, 1301, 0, "200", "1301", "0", "0429", " 429", "\u0664\u0662\u0669", "4" * 5000], +) +def test_a_carried_status_that_is_not_an_error_status_falls_back_to_the_default(carried_status: object): + """True is an int in Python but not an HTTP status, and answering a 200 or a 1301 either reads + as success or makes uvicorn drop the connection.""" class _Carrier(Exception): status_code = carried_status @@ -122,6 +125,22 @@ def test_a_non_int_carried_status_falls_back_to_the_default(carried_status: obje assert error_status_code(_Carrier("boom"), 500) == 500 +def test_a_status_carried_as_a_decimal_string_is_answered_as_that_status(): + """Provider stream-error chunks pass the vendor's code through as a string ``status_code``, + and the shared handler used to answer it as is, so dropping it turned a rate limit into a 500.""" + + class _Carrier(Exception): + status_code = "429" + + assert error_status_code(_Carrier("slow down"), 500) == 429 + assert proxy_exception_for(_Carrier("slow down"), default_status_code=500).to_dict() == { + "message": "slow down", + "type": "rate_limit_error", + "param": None, + "code": "429", + } + + def test_a_proxy_exception_keeps_the_status_it_was_raised_with(): """ProxyException stores its status as the string ``code`` rather than ``status_code``, so a route tail that rewraps one used to answer a 4xx rejection as a 500.""" @@ -130,8 +149,8 @@ def test_a_proxy_exception_keeps_the_status_it_was_raised_with(): assert error_status_code(rejection, 500) == 400 -@pytest.mark.parametrize("carried_code", [None, "None", "", "rate_limited", "4xx", 404]) -def test_a_code_that_is_not_a_decimal_string_falls_back_to_the_default(carried_code: object): +@pytest.mark.parametrize("carried_code", [None, "None", "", "rate_limited", "4xx", 404, "200", "1301"]) +def test_a_code_that_is_not_a_stringified_error_status_falls_back_to_the_default(carried_code: object): """Only ProxyException's stringified status is a status; ``code`` on anything else (OpenAI's ``invalid_api_key``, a stray int) says nothing about the HTTP answer.""" diff --git a/tests/unit/proxy/credential_endpoints/test_endpoints.py b/tests/unit/proxy/credential_endpoints/test_endpoints.py index 47dd6d561b3..3d8e92988d9 100644 --- a/tests/unit/proxy/credential_endpoints/test_endpoints.py +++ b/tests/unit/proxy/credential_endpoints/test_endpoints.py @@ -3277,6 +3277,23 @@ class TestCredentialDisplayName: assert listed["openai-prod"]["credential_values"]["api_key"] != "sk-old" +def test_looking_up_a_missing_credential_is_labelled_a_client_error_not_a_gateway_failure( + restore_credential_list: None, +): + """Naming a credential that does not exist is the caller's own mistake, and the shared + handler used to label every one of these internal_server_error, so an SDK branching on + the type read a 404 it can never retry past as a gateway outage and backed off forever.""" + response: Final = _call_as("GET", "/credentials/by_name/definitely-not-there") + + assert response.status_code == 404, f"lookup answered {response.status_code}: {response.text}" + assert response.json()["error"] == { + "message": "Credential not found. Got credential name: definitely-not-there", + "type": "invalid_request_error", + "param": None, + "code": "404", + } + + def test_get_credential_by_model_masks_the_deployment_credentials(monkeypatch): import asyncio diff --git a/tests/unit/proxy/guardrails/test_guardrail_endpoints.py b/tests/unit/proxy/guardrails/test_guardrail_endpoints.py index 6807a4fb569..7d912478877 100644 --- a/tests/unit/proxy/guardrails/test_guardrail_endpoints.py +++ b/tests/unit/proxy/guardrails/test_guardrail_endpoints.py @@ -13,7 +13,7 @@ import yaml from fastapi import HTTPException from litellm.integrations.custom_guardrail import CustomGuardrail -from litellm.proxy._types import LitellmUserRoles, UserAPIKeyAuth +from litellm.proxy._types import LitellmUserRoles, ProxyException, UserAPIKeyAuth from litellm.proxy.guardrails.guardrail_endpoints import ( CreateGuardrailRequest, ListGuardrailsResponse, @@ -2198,6 +2198,39 @@ def _patch_apply_guardrail_env(mocker, guardrail_result): return mock_guardrail +@pytest.mark.asyncio +@pytest.mark.parametrize( + ("vendor_status", "expected_type"), + [(401, "authentication_error"), (403, "permission_error")], +) +async def test_a_guardrail_vendors_rejection_is_labelled_the_way_chat_labels_it( + mocker, vendor_status: int, expected_type: str +): + """A vendor refusing the guardrail's own credentials is not the caller's key being rejected, + so it must not read as the proxy's auth_error the way a 401 the proxy raises does.""" + mock_guardrail: Final = _patch_apply_guardrail_env(mocker, None) + mock_guardrail.apply_guardrail.side_effect = HTTPException( + status_code=vendor_status, detail="vendor refused the request" + ) + mocker.patch( + "litellm.proxy.proxy_server.proxy_logging_obj.post_call_failure_hook", AsyncMock(return_value=None) + ) + + with pytest.raises(ProxyException) as raised: + await apply_guardrail( + fastapi_request=mocker.Mock(), + request=ApplyGuardrailRequest(guardrail_name="test-guardrail", text="hello"), + user_api_key_dict=UserAPIKeyAuth(), + ) + + assert raised.value.to_dict() == { + "message": "vendor refused the request", + "type": expected_type, + "param": None, + "code": str(vendor_status), + } + + @pytest.mark.asyncio async def test_apply_guardrail_forwards_metadata_to_guardrail(mocker): """Client-supplied metadata must reach apply_guardrail via request_data so diff --git a/tests/unit/proxy/management_endpoints/test_key_management_endpoints.py b/tests/unit/proxy/management_endpoints/test_key_management_endpoints.py index a0c32f2235f..62dd367b8e5 100644 --- a/tests/unit/proxy/management_endpoints/test_key_management_endpoints.py +++ b/tests/unit/proxy/management_endpoints/test_key_management_endpoints.py @@ -18750,7 +18750,7 @@ async def test_list_keys_rejects_invalid_expires(): expires="expred", ) - assert exc_info.value.code == "400" + assert (exc_info.value.code, exc_info.value.type) == ("400", "invalid_request_error") assert "Invalid expires value" in str(exc_info.value.message) diff --git a/tests/unit/proxy/utils/helpers/test_error_helpers.py b/tests/unit/proxy/utils/helpers/test_error_helpers.py index df399c8b1d2..b20857dbdaf 100644 --- a/tests/unit/proxy/utils/helpers/test_error_helpers.py +++ b/tests/unit/proxy/utils/helpers/test_error_helpers.py @@ -1,13 +1,20 @@ import asyncio import json +from typing import Final from unittest.mock import AsyncMock, MagicMock +import httpx +import openai import pytest from fastapi import HTTPException from prisma.errors import DataError +import litellm from litellm.proxy._types import ProxyErrorTypes, ProxyException +from litellm.proxy.common_request_processing import proxy_exception_from_route_error +from litellm.proxy.common_utils.openai_error_payload import proxy_exception_for from litellm.proxy.utils import get_error_message_str, handle_exception_on_proxy +from litellm.types.router import RouterErrors def normalize(value): @@ -96,7 +103,7 @@ def test_handle_exception_on_proxy_happy_path_http_exception(): assert snapshot == { "is_proxy_exception": True, "message": "forbidden", - "type": ProxyErrorTypes.internal_server_error.value, + "type": "auth_error", "code": "403", } @@ -156,7 +163,7 @@ def test_handle_exception_on_proxy_uses_attached_status_code_when_present(): assert snapshot == { "code": "418", "message": "teapot", - "type": ProxyErrorTypes.internal_server_error.value, + "type": "invalid_request_error", } @@ -176,6 +183,176 @@ def test_handle_exception_on_proxy_error_path_none_input_wraps_as_500(): } +@pytest.mark.parametrize( + "status_code, expected_type", + [ + (400, "invalid_request_error"), + (401, "auth_error"), + (403, "auth_error"), + (404, "invalid_request_error"), + (408, "invalid_request_error"), + (409, "invalid_request_error"), + (422, "invalid_request_error"), + (429, "rate_limit_error"), + (500, ProxyErrorTypes.internal_server_error.value), + (503, ProxyErrorTypes.internal_server_error.value), + ], +) +def test_handle_exception_on_proxy_labels_an_http_exception_by_the_status_it_answers( + status_code: int, expected_type: str +): + """A management route that rejects a caller's own argument answers 400, and calling that + internal_server_error tells the caller's retry loop the gateway is down when nothing but + their request is wrong.""" + result: Final = handle_exception_on_proxy(HTTPException(status_code=status_code, detail="boom")) + + assert (result.code, result.type) == (str(status_code), expected_type) + + +def _exception_carrying_status(carried: int | None, message: str = "boom") -> Exception: + class _CarriedStatus(Exception): + status_code = carried + + return _CarriedStatus(message) + + +@pytest.mark.parametrize( + "status_code, expected_type", + [ + (400, "invalid_request_error"), + (404, "invalid_request_error"), + (408, "invalid_request_error"), + (429, "rate_limit_error"), + (500, ProxyErrorTypes.internal_server_error.value), + ], +) +def test_handle_exception_on_proxy_labels_a_carried_status_the_same_way(status_code: int, expected_type: str): + """Routes also reach this handler with plain exceptions carrying their own status, and + those must be named by that status rather than all landing on the 500 label.""" + result: Final = handle_exception_on_proxy(_exception_carrying_status(status_code)) + + assert (result.code, result.type) == (str(status_code), expected_type) + + +def test_handle_exception_on_proxy_keeps_the_type_the_exception_already_names(): + """A rate limit litellm raised already names itself, and overwriting that with the status + map's guess would drop what the client branches on.""" + result: Final = handle_exception_on_proxy( + litellm.RateLimitError(message="slow down", llm_provider="openai", model="gpt-4o") + ) + + assert (result.code, result.type) == ("429", "throttling_error") + + +@pytest.mark.parametrize( + "exc", + [ + openai.InternalServerError( + message="upstream down", + response=httpx.Response(500, request=httpx.Request("GET", "https://api.openai.com/v1/batches/b")), + body={"type": "server_error"}, + ), + litellm.Timeout(message="took too long", model="gpt-6.1-sol", llm_provider="openai"), + litellm.RateLimitError(message="slow down", llm_provider="openai", model="gpt-6.1-sol"), + litellm.AuthenticationError(message="bad key", llm_provider="openai", model="gpt-6.1-sol"), + litellm.BadRequestError(message="bad input", model="gpt-6.1-sol", llm_provider="openai"), + HTTPException(status_code=404, detail="no such batch"), + ], + ids=["provider_5xx_naming_its_own_type", "timeout", "rate_limit", "bad_key", "bad_request", "http_404"], +) +def test_handle_exception_on_proxy_labels_a_failure_the_way_the_chat_route_does(exc: Exception): + """Batch, file, and management routes answer through this handler while chat answers through the route + error path, and one client calling both must read the same type for the same failure.""" + shared: Final = handle_exception_on_proxy(exc) + chat: Final = proxy_exception_from_route_error(exc) + + assert (shared.code, shared.type) == (chat.code, chat.type) + + +@pytest.mark.parametrize( + "exc", + [ + HTTPException(status_code=400, detail="bad budget_duration"), + HTTPException(status_code=401, detail="not allowed"), + HTTPException(status_code=403, detail="admin only"), + HTTPException(status_code=404, detail="no such team"), + litellm.AuthenticationError(message="bad key", llm_provider="openai", model="gpt-6.1-sol"), + ValueError(f"{RouterErrors.no_deployments_with_tag_routing.value}=['paid']"), + _exception_carrying_status(409, "duplicate credential"), + ], + ids=["http_400", "http_401", "http_403", "http_404", "provider_401", "tag_denial", "carried_409"], +) +def test_handle_exception_on_proxy_labels_a_failure_the_way_route_catch_alls_do(exc: Exception): + """Management routes answer some failures through their own catch-all and the rest through this handler, + so a client reading /key/list and /key/generate must see the same type and status for the same failure.""" + shared: Final = handle_exception_on_proxy(exc) + catch_all: Final = proxy_exception_for(exc, default_status_code=500) + + assert (shared.code, shared.type, shared.message) == (catch_all.code, catch_all.type, catch_all.message) + + +@pytest.mark.parametrize( + "message, expected", + [ + (f"{RouterErrors.no_deployments_with_provider_budget_routing.value}, try later", ("429", "rate_limit_error")), + (f"{RouterErrors.no_deployments_with_tag_routing.value}=['paid']", ("401", "auth_error")), + ], + ids=["no_deployment_left", "no_deployment_for_tags"], +) +def test_handle_exception_on_proxy_names_a_router_failure_by_the_status_it_answers( + message: str, expected: tuple[str, str] +): + """A router failure that carries no status answers the 429 or 401 its message maps to, and a 429 labelled + internal_server_error reads as an outage to a client that branches on the type.""" + result: Final = handle_exception_on_proxy(ValueError(message)) + + assert (result.code, result.type) == expected + + +def test_handle_exception_on_proxy_keeps_the_path_a_rejected_request_names(): + """A provider rejecting a batch input names the caller's own bucket path, and redacting the /tmp part of it + leaves the caller nothing to fix their request with.""" + result: Final = handle_exception_on_proxy( + litellm.BadRequestError( + message="Invalid input_file_id gs://my-bucket/tmp/batch_input.jsonl", + model="gemini-3.1-pro-preview", + llm_provider="vertex_ai", + ) + ) + + assert result.code == "400" + assert "gs://my-bucket/tmp/batch_input.jsonl" in result.message + + +def test_handle_exception_on_proxy_keeps_the_message_an_exception_with_a_status_carries(): + """Main sends a status-bearing failure's message verbatim, and relabeling its type must not change that text.""" + result: Final = handle_exception_on_proxy(_exception_carrying_status(503, "Cannot connect to host 10.12.4.7:8443")) + + assert (result.code, result.message) == ("503", "Cannot connect to host 10.12.4.7:8443") + + +@pytest.mark.parametrize( + "message", + [ + "boom", + "Cannot connect to host 10.12.4.7:8443", + "Cannot connect to host guardrail.internal.svc.cluster.local:443", + "Cannot connect to host [fd00::12]:8443", + ], + ids=["plain", "ipv4", "cluster_dns", "ipv6"], +) +def test_handle_exception_on_proxy_answers_a_generic_500_for_an_exception_whose_status_is_none(message: str): + """A guardrail whose upstream refuses the connection raises with status_code=None, which crashed the handler + into a 503 blaming the auth database, and its text names private hosts no virtual key was ever sent.""" + result: Final = handle_exception_on_proxy(_exception_carrying_status(None, message)) + + assert (result.code, result.type, result.message) == ( + "500", + ProxyErrorTypes.internal_server_error.value, + "Internal server error", + ) + + @pytest.mark.parametrize( "exc", [ From 160a0b72cff4dbbc41bd5de1130966a207f0c17d Mon Sep 17 00:00:00 2001 From: simonp9 <349053457+simonp9@users.noreply.github.com> Date: Fri, 30 Oct 2026 07:34:10 -0700 Subject: [PATCH 03/18] fix(mcp): bill both calls of a streamed chat completions MCP auto-execute (#48899) * fix(mcp): bill both rounds of a streamed chat completions MCP auto-execute A streamed /v1/chat/completions request with an MCP tool set to require_approval "never" runs two model calls on one logging object, and the final-stream dedupe kept only the first round's spend row. The final usage chunk also showed only the second round. The MCP stream iterator now holds the first round's assembled response, prices it, and logs one success row that carries the usage and cost of both rounds, with the fixed margin charged once. The final usage chunk shows the same summed usage. A single-round stream still logs its one row when the answer finishes. When the follow-up round fails, the first round is logged from a copy of the logging object taken before the follow-up opened, and the failure row carries only the follow-up's partial usage with no second fixed margin. With proxy post-call guardrails armed, only (assembled, cache_hit) tuples are parked for the proxy, and the deferred log is armed only while the iterator reads, so the proxy's orphaned-callback cleanup never sees it after the Router's first-chunk read * fix(mcp): send the usage chunk of a streamed chat MCP request that ends in one round When a streamed chat completions MCP request ends in its first round (the model answers without a tool call, or no tool results come back), the iterator drained the inner stream to fire its spend log and dropped the usage chunk that stream produced, so a client that asked for include_usage got no usage chunk at all. The drained usage chunk is now held and sent to the client after the final content chunk and before [DONE], the same place a plain stream sends it, so the last usage chunk matches the billed row. A two-round request still drops the first round's usage chunk and sends the merged one from the follow-up, and a stream without include_usage still sends none * fix(mcp): hold round one's log for the guardrail, bill the chunk cost once, and keep round one's id When the follow-up call of a streamed chat MCP auto-execute fails, round one's success log now goes through the proxy's deferred guardrail release instead of straight to the handlers, so a post-call guardrail runs on it and the row carries it. Without a proxy release it logs directly as before The last usage chunk now takes its cost from the combined row's billed total, so a fixed cost margin is counted once in the chunk, as it is on the row. The follow-up's usage chunk is held until the follow-up ends to make that possible The combined row now keeps round one's id, the stream's first chunk id, which matches how a single-round stream is logged * fix(mcp): bill every chat MCP stream round once on failure, hang-up, and sync drop When the follow-up call fails, round 1's success row is now logged on a copy of the logging object that does not share the request's guardrail results, so a post-call guardrail that reports on the failed request counts once instead of on both rows A client hang-up during the follow-up now bills round 1 plus the follow-up's partial answer on one success row with round 1's id and the fixed margin counted once, the same way a plain stream's hang-up bills its partial answer The last usage chunk now carries round 1's usage and cost when only round 1 is billed, and a synchronous reader that drops the stream mid-answer bills round 1 on the loop it read with instead of discarding it The deferred release's logged_on argument is typed and every release closure accepts it, and the parked dispatch constants and protocols both MCP billing paths use now live in one module. The unit tests wait on a logging signal instead of a fixed delay and check the billed row through the proxy stream hook instead of private callback state * test(mcp): drive the unassembled follow-up through an upstream audio delta that is not base64 The test replaced litellm.stream_chunk_builder with a stub that raises. Round 2 now streams an audio delta whose data is not valid base64, so the real builder raises, the stream parks only best-effort usage, and the client's last usage chunk must carry round 1's tokens and cost * fix(mcp): count the fixed margin once in the combined stream row's usage cost * fix(mcp): bill round 2's partial answer when a client hangs up on a routed stream On the proxy the Router wraps the MCP stream, and when a client hangs up during round 2 the router's own cleanup closes that stream before the proxy's disconnect cleanup runs. The close logged round 1 alone, so round 2's partial answer was never billed. The MCP stream now bills round 1 plus round 2's partial answer as soon as the hang-up's cancellation reaches its read, before anything closes it * fix(mcp): count the follow-up's own messages in round 2's partial prompt The follow-up stream shares the request's logger, so it took the client's original messages as its prompt. When round 2 ended early, from a hang-up or a broken upstream, its partial usage counted only those messages and left out the assistant tool call and the tool results the follow-up actually sent. The follow-up stream now counts the messages the follow-up call sent * fix(mcp): bill round 2's partial when a routed stream's client hangs up between reads On a proxy the stream the disconnect cleanup holds is the router's fallback wrapper, so the cleanup never found the MCP stream's in-flight billing and released round 1 alone when the client hung up while the proxy sat at a yield. The round billing now registers its in-flight biller on the request's logging object, next to the parked release, and the cleanup looks it up there, so any wrapper chain bills round 1 plus round 2's partial once * fix(mcp): send the MCP stream usage chunk under the billed row id and count the partial off the loop A streamed chat completion that runs a LiteLLM-hosted MCP tool bills both model calls on one row under round 1's chatcmpl id, but the usage chunk it sent the client was round 2's held chunk and kept round 2's id, so a caller could not join the usage chunk to its row. The usage chunk now carries round 1's id in every path that sends it, with or without a deferred post-call guardrail, while its usage and cost and every content chunk's id stay as they were When the client hangs up during round 2, assembling round 2's partial answer counts the follow-up messages, tool result included, with the local tokenizer. That ran on the event loop and stalled every other request on the worker for about a second on a large tool result. The assembly now runs on a worker thread through asyncify from both the reader's cancellation and the proxy's disconnect cleanup, shielded from cancellation, and round 1 is checked again after the await so a release that happened meanwhile is never billed twice * test(mcp): wait on the loop with a hang guard instead of a 300 ms deadline * fix(mcp): give a proxy-rejected follow-up's failure row round 1's own cost breakdown * refactor(streaming): fold the partial response's empty guards into one check --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> --- litellm/litellm_core_utils/litellm_logging.py | 22 + .../litellm_core_utils/streaming_handler.py | 35 +- litellm/proxy/common_request_processing.py | 45 +- litellm/proxy/utils.py | 3 + .../responses/mcp/chat_completions_handler.py | 101 +- .../mcp/chat_stream_round_billing.py | 281 +++++ .../responses/mcp/parked_stream_dispatch.py | 72 ++ litellm/responses/mcp/stream_round_billing.py | 58 +- .../integration/_support/hook_count_probe.py | 18 + .../integration/mcp/test_mcp_llm_endpoints.py | 261 +++- .../proxy/test_common_request_processing.py | 28 + .../mcp/test_chat_completions_handler.py | 1050 ++++++++++++++++- .../mcp/test_parked_stream_dispatch.py | 17 + 13 files changed, 1894 insertions(+), 97 deletions(-) create mode 100644 litellm/responses/mcp/chat_stream_round_billing.py create mode 100644 litellm/responses/mcp/parked_stream_dispatch.py create mode 100644 tests/unit/responses/mcp/test_parked_stream_dispatch.py diff --git a/litellm/litellm_core_utils/litellm_logging.py b/litellm/litellm_core_utils/litellm_logging.py index 4dd33fa5565..8bfb0b60693 100644 --- a/litellm/litellm_core_utils/litellm_logging.py +++ b/litellm/litellm_core_utils/litellm_logging.py @@ -568,6 +568,11 @@ def _resolve_mantle_region_for_cost( _PER_CALL_LITELLM_PARAM_DICTS: Final = frozenset({"metadata", "litellm_metadata"}) _LITELLM_PARAMS: Final = TypeAdapter(dict[str, object]) _REDACTED_RESULT: Final = TypeAdapter(object) +_GUARDRAIL_INFORMATION_KEY: Final = "standard_logging_guardrail_information" + + +def _without_guardrail_information(metadata: Mapping[str, object]) -> Mapping[str, object]: + return {key: value for key, value in metadata.items() if key != _GUARDRAIL_INFORMATION_KEY} def _provider_response_id(source: object) -> str | None: @@ -2566,6 +2571,23 @@ class Logging(LiteLLMLoggingBaseClass): attempt.model_call_details = dict(self.model_call_details) return attempt + def guardrail_scoped_copy(self) -> "Logging": + request_params: Final = _LITELLM_PARAMS.validate_python( + self.litellm_params or {} # pyright: ignore[reportUnknownMemberType] # Logging.litellm_params is untyped + ) + litellm_params: Final = { + **request_params, + **{ + key: _without_guardrail_information(_METADATA_MAPPING.validate_python(value)) + for key, value in request_params.items() + if key in _PER_CALL_LITELLM_PARAM_DICTS and isinstance(value, dict) + }, + } + scoped: Final = copy.copy(self) + scoped.litellm_params = litellm_params + scoped.model_call_details = {**self.model_call_details, "litellm_params": litellm_params} + return scoped + def dispatch_attempt_failure_handlers( self, exception: Exception, traceback_exception: str ) -> Coroutine[None, None, None]: diff --git a/litellm/litellm_core_utils/streaming_handler.py b/litellm/litellm_core_utils/streaming_handler.py index 682b2a2f538..098a404ea14 100644 --- a/litellm/litellm_core_utils/streaming_handler.py +++ b/litellm/litellm_core_utils/streaming_handler.py @@ -2486,6 +2486,25 @@ class CustomStreamWrapper: def bill_failure_after_earlier_rounds(self, earlier: "BilledRoundsSpend") -> None: self._earlier_rounds_spend = earlier + def partial_response(self) -> "ModelResponse | TextCompletionResponse | None": + partial: Final = ( + litellm.stream_chunk_builder( + chunks=self.chunks, + messages=self.messages if isinstance(self.messages, list) else None, + logging_obj=self.logging_obj, + count_prompt_tokens=self.count_prompt_tokens, + ) + if self.chunks + else None + ) + usage: Final[object] = getattr(partial, "usage", None) + if partial is None or not isinstance(usage, Usage): + return None + if self.model: + partial.model = self.model + backfill_missing_cache_usage_fields(usage) + return partial + def _record_partial_usage_for_failure(self) -> None: """ A stream that breaks mid-flight still billed the provider for the chunks @@ -2495,23 +2514,15 @@ class CustomStreamWrapper: moves it into terminal failure usage before opening the next attempt, so a recovered attempt logs its own usage. """ - if self.logging_obj is None or not self.chunks: + if self.logging_obj is None: return try: - partial_response: Final = litellm.stream_chunk_builder( - chunks=self.chunks, - messages=self.messages if isinstance(self.messages, list) else None, - logging_obj=self.logging_obj, - count_prompt_tokens=self.count_prompt_tokens, - ) + partial_response: Final = self.partial_response() if partial_response is None: return - usage: Final = cast(Usage | None, getattr(partial_response, "usage", None)) - if usage is None: + usage: Final[object] = getattr(partial_response, "usage", None) + if not isinstance(usage, Usage): return - if self.model: - partial_response.model = self.model - backfill_missing_cache_usage_fields(usage) partial_cost: Final = self.logging_obj.response_cost_calculator(result=partial_response) or 0.0 if self._earlier_rounds_spend is None: self.logging_obj.model_call_details["combined_usage_object"] = usage diff --git a/litellm/proxy/common_request_processing.py b/litellm/proxy/common_request_processing.py index a79d8fe29db..48eb58adaf2 100644 --- a/litellm/proxy/common_request_processing.py +++ b/litellm/proxy/common_request_processing.py @@ -140,6 +140,7 @@ from litellm.proxy.utils import ( # noqa: F401 # legacy module exports _check_and_merge_model_level_guardrails, # pyright: ignore[reportPrivateUsage,reportUnusedImport] # backwards-compatible package export check_and_merge_model_level_guardrails, ) +from litellm.responses.mcp.parked_stream_dispatch import in_flight_round_biller, parked_success from litellm.responses.streaming_iterator import partial_responses_incomplete_event_on_disconnect, spawn_logging_task from litellm.router import Router from litellm.router_api import RoutingContext @@ -535,6 +536,16 @@ async def _bill_partial_streamed_spend_on_disconnect(request_data: dict, respons return await _dispatch_disconnect_billing_event(logging_obj, partial_response) +async def _bill_in_flight_round_on_disconnect(request_data: Mapping[str, object]) -> None: + bill_in_flight_round: Final = in_flight_round_biller(request_data.get("litellm_logging_obj")) + if bill_in_flight_round is None: + return + try: + await bill_in_flight_round() + except Exception as e: # noqa: BLE001 # disconnect billing is best-effort; never break stream teardown + verbose_proxy_logger.debug("Failed to bill the in-flight round after client disconnect: %s", e) + + async def _dispatch_disconnect_billing_event(logging_obj: LiteLLMLoggingObj, partial_response: object) -> bool: try: await logging_obj.dispatch_success_handlers( @@ -2083,6 +2094,13 @@ class _ClientStreamFailureReport: ) +def _metadata_logged_on(logged_on: LiteLLMLoggingObj) -> Mapping[str, object]: + logged_params: Final = object_mapping(logged_on.model_call_details.get("litellm_params")) or MappingProxyType({}) + return MappingProxyType( + {key: logged_params[key] for key in ("metadata", "litellm_metadata") if key in logged_params} + ) + + class ProxyBaseLLMRequestProcessing: def __init__(self, data: dict, served_model: str | None = None): self.data = data @@ -4151,11 +4169,21 @@ class ProxyBaseLLMRequestProcessing: _captured_user_api_key_dict: Final = user_api_key_dict _captured_logging_obj: Final = logging_obj - async def _on_deferred_stream_complete(assembled_response: object, cache_hit: object) -> None: + async def _on_deferred_stream_complete( + assembled_response: object, cache_hit: object, logged_on: LiteLLMLoggingObj | None = None + ) -> None: await ProxyBaseLLMRequestProcessing._run_deferred_stream_guardrails( - captured_data=_captured_data, + captured_data=_captured_data + if logged_on is None + else { + **_captured_data, + **{ + key: value for key, value in _metadata_logged_on(logged_on).items() if key in _captured_data + }, + "litellm_logging_obj": logged_on, + }, captured_user_api_key_dict=_captured_user_api_key_dict, - captured_logging_obj=_captured_logging_obj, + captured_logging_obj=_captured_logging_obj if logged_on is None else logged_on, assembled_response=assembled_response, cache_hit=cache_hit, ) @@ -4180,14 +4208,8 @@ class ProxyBaseLLMRequestProcessing: GLOBAL_LOGGING_WORKER.ensure_initialized_and_enqueue(async_coroutine=logging_coroutine) case (logging_coroutine,) if asyncio.iscoroutine(logging_coroutine): GLOBAL_LOGGING_WORKER.ensure_initialized_and_enqueue(async_coroutine=logging_coroutine) - case (assembled_response, cache_hit): - await _as_success_dispatcher(_captured_native_logging_obj).dispatch_success_handlers( - assembled_response, - cache_hit=cache_hit, - start_time=None, - end_time=None, - prefer_async_handlers=True, - ) + case _ if (success := parked_success(args)) is not None: + await success.dispatch(_captured_native_logging_obj) case _: verbose_proxy_logger.error( "Deferred stream logging dropped: unexpected stored args shape %s", @@ -4524,6 +4546,7 @@ class ProxyBaseLLMRequestProcessing: client_disconnected, ) if recorded_client_disconnect: + await _bill_in_flight_round_on_disconnect(_REQUEST_FIELDS.validate_python(request_data)) deferred_stream_logging_armed: Final = _deferred_stream_logging_is_armed(request_data) ProxyLogging.fire_deferred_stream_logging(request_data) # A disconnect-time success event (the deferred-guardrail flush diff --git a/litellm/proxy/utils.py b/litellm/proxy/utils.py index 74e16628bc8..ddaa8d2e69d 100644 --- a/litellm/proxy/utils.py +++ b/litellm/proxy/utils.py @@ -91,6 +91,7 @@ from litellm.proxy.common_utils.openai_error_payload import ( ) from litellm.proxy.spend_tracking.spend_log_error_logger import spend_log_error from litellm.responses.litellm_completion_transformation.compaction_items import messages_since_newest_compaction +from litellm.responses.mcp.parked_stream_dispatch import held_round_spend from litellm.types.guardrails import GuardrailEventHooks from litellm.types.proxy.model_listing import ModelInfoResponse from litellm.types.utils import MCP_GUARDRAIL_CALL_TYPES, CallTypes, CallTypesLiteral, ModelInfo, Usage @@ -4943,6 +4944,8 @@ class ProxyLogging: if not isinstance(error, _PROXY_ONLY_LLM_API_ERRORS): return False match _args: + case (ModelResponse(), *_) if (held := held_round_spend(logging_obj)) is not None: + logging_obj.record_billed_rounds_for_failure(held) case (ModelResponse() as assembled, *_): logging_obj.record_assembled_response_for_failure(assembled) case (BilledRoundsDispatch() as billed,): diff --git a/litellm/responses/mcp/chat_completions_handler.py b/litellm/responses/mcp/chat_completions_handler.py index cd072825ef5..1e07b768f62 100644 --- a/litellm/responses/mcp/chat_completions_handler.py +++ b/litellm/responses/mcp/chat_completions_handler.py @@ -1,15 +1,18 @@ """Helpers for handling MCP-aware `/chat/completions` requests.""" +import asyncio import logging from types import MappingProxyType from typing import TYPE_CHECKING, Final, cast +import anyio from pydantic import TypeAdapter from typing_extensions import TypedDict, Unpack from litellm._internal_context import internal_sub_call from litellm.litellm_core_utils.hidden_params import HIDDEN_PARAMS_ATTR, set_hidden_params from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj +from litellm.responses.mcp.chat_stream_round_billing import ChatStreamRoundBilling, engaged_round_billing from litellm.responses.mcp.litellm_proxy_mcp_handler import ( LiteLLM_Proxy_MCP_Handler, ) @@ -19,7 +22,7 @@ from litellm.responses.mcp.round_billing import ( first_round_billed_on_failure, summed_cost_breakdown, ) -from litellm.types.utils import Message, ModelResponse +from litellm.types.utils import Message, ModelResponse, Usage from litellm.utils import CustomStreamWrapper if TYPE_CHECKING: @@ -258,10 +261,20 @@ async def acompletion_with_mcp( self.follow_up_stream = None self.follow_up_iterator = None self.follow_up_exhausted = False + self._round_billing: ChatStreamRoundBilling | None = engaged_round_billing(initial_stream.logging_obj) + self._unsent_usage_chunk: ModelResponseStream | None = None def __aiter__(self): return self + def __del__(self) -> None: + self.release_abandoned() + + def release_abandoned(self) -> None: + round_billing: Final[object] = getattr(self, "_round_billing", None) + if isinstance(round_billing, ChatStreamRoundBilling): + round_billing.release_abandoned() + def _add_mcp_list_tools_to_chunk(self, chunk: ModelResponseStream) -> ModelResponseStream: """Add mcp_list_tools to the first chunk.""" from litellm.types.utils import ( @@ -321,18 +334,54 @@ async def acompletion_with_mcp( return chunk - async def _drain_inner_stream(self): + async def _drain_inner_stream(self) -> tuple[ModelResponseStream, ...]: try: - while True: - await self._stream_iterator.__anext__() - except StopAsyncIteration: - pass + return tuple([chunk async for chunk in initial_stream]) except Exception: logging.getLogger("LiteLLM").exception( "Error draining inner MCP stream after final chunk; spend logging may be incomplete" ) + return () - async def __anext__(self): + async def _complete_first_round(self) -> ModelResponseStream | None: + drained: Final = await self._drain_inner_stream() + if self._round_billing is not None: + self._round_billing.complete_first_round() + usage_chunks: Final = tuple( + chunk for chunk in drained if isinstance(getattr(chunk, "usage", None), Usage) + ) + return usage_chunks[-1] if usage_chunks else None + + async def _follow_up_or_release_first_round(self, usage_chunk: ModelResponseStream | None) -> None: + if self.tool_results and self.complete_response: + await self._prepare_follow_up_call() + if self.follow_up_stream is not None: + return + self._unsent_usage_chunk = usage_chunk + if self._round_billing is not None: + self._round_billing.finish() + + async def __anext__(self) -> ModelResponseStream: + round_billing: Final = self._round_billing + if round_billing is None: + return await self._next_chunk() + with round_billing.armed_while_reading(): + try: + return await self._next_chunk() + except StopAsyncIteration: + usage_chunk: Final = round_billing.finish() + if usage_chunk is None: + raise + return usage_chunk + except asyncio.CancelledError: + with anyio.CancelScope(shield=True): + await round_billing.bill_in_flight_round() + raise + except Exception: + round_billing.fail() + raise + + async def _next_chunk(self) -> ModelResponseStream: # Phase 1: Collect and yield initial stream chunks if not self.stream_exhausted: # Get the iterator from the stream wrapper @@ -362,29 +411,22 @@ async def acompletion_with_mcp( if is_final: self.stream_exhausted = True + usage_chunk: Final = await self._complete_first_round() await self._process_tool_calls() chunk = self._add_mcp_tool_metadata_to_final_chunk(chunk) - if self.tool_results and self.complete_response: - await self._prepare_follow_up_call() - # Drain inner stream so CustomStreamWrapper fires its - # end-of-stream handler (dispatch_success_handlers → - # _ProxyDBLogger → LiteLLM_SpendLogs). The CSW may - # yield one usage chunk before raising StopAsyncIteration. - await self._drain_inner_stream() + await self._follow_up_or_release_first_round(usage_chunk) return chunk except StopAsyncIteration: self.stream_exhausted = True + drained_usage_chunk: Final = await self._complete_first_round() # Process tool calls after stream is exhausted await self._process_tool_calls() # If we have chunks, yield the final one with metadata if self.collected_chunks: final_chunk = self.collected_chunks[-1] final_chunk = self._add_mcp_tool_metadata_to_final_chunk(final_chunk) - # If we have tool results, prepare follow-up call - if self.tool_results and self.complete_response: - await self._prepare_follow_up_call() - await self._drain_inner_stream() + await self._follow_up_or_release_first_round(drained_usage_chunk) return final_chunk # Phase 2: Yield follow-up stream chunks if available @@ -400,7 +442,6 @@ async def acompletion_with_mcp( from litellm._logging import verbose_logger verbose_logger.debug("Follow-up chunk yielded: %s", chunk) - return chunk except StopAsyncIteration: self.follow_up_exhausted = True from litellm._logging import verbose_logger @@ -408,6 +449,10 @@ async def acompletion_with_mcp( verbose_logger.debug("Follow-up stream exhausted") # After follow-up stream is exhausted, check if we need to raise StopAsyncIteration raise StopAsyncIteration + client_chunk: Final = ( + chunk if self._round_billing is None else self._round_billing.client_chunk(chunk) + ) + return await self._next_chunk() if client_chunk is None else client_chunk # If we're here and follow_up_stream is None but we expected it, log a warning if ( @@ -420,7 +465,11 @@ async def acompletion_with_mcp( verbose_logger.warning("Follow-up stream was not created despite having tool results") - raise StopAsyncIteration + unsent_usage_chunk: Final = self._unsent_usage_chunk + if unsent_usage_chunk is None: + raise StopAsyncIteration + self._unsent_usage_chunk = None + return unsent_usage_chunk async def _process_tool_calls(self): """Process tool calls after streaming completes.""" @@ -488,10 +537,16 @@ async def acompletion_with_mcp( # This ensures the patch works correctly in tests import litellm + earlier_rounds: Final = None if self._round_billing is None else self._round_billing.bank_first_round() follow_up_response: Final = await litellm.acompletion(**follow_up_call_args) # Ensure follow-up response is a CustomStreamWrapper if isinstance(follow_up_response, CustomStreamWrapper): + if earlier_rounds is not None: + follow_up_response.bill_failure_after_earlier_rounds(earlier_rounds) + follow_up_response.messages = follow_up_messages + if self._round_billing is not None: + self._round_billing.follow_up_started(follow_up_response.partial_response) self.follow_up_stream = follow_up_response from litellm._logging import verbose_logger @@ -548,11 +603,13 @@ async def acompletion_with_mcp( def __aiter__(self): return self._custom_iterator + async def aclose(self) -> None: + self._custom_iterator.release_abandoned() + await super().aclose() + def __iter__(self): # For synchronous iteration, create a sync wrapper if self._sync_iterator is None: - import asyncio - try: self._sync_loop = asyncio.get_event_loop() except RuntimeError: diff --git a/litellm/responses/mcp/chat_stream_round_billing.py b/litellm/responses/mcp/chat_stream_round_billing.py new file mode 100644 index 00000000000..370de663606 --- /dev/null +++ b/litellm/responses/mcp/chat_stream_round_billing.py @@ -0,0 +1,281 @@ +import asyncio +from collections.abc import Callable, Generator +from contextlib import contextmanager +from dataclasses import dataclass, replace +from types import MappingProxyType +from typing import Final, TypeAlias + +from litellm.litellm_core_utils.asyncify import asyncify, is_event_loop_running +from litellm.litellm_core_utils.hidden_params import get_hidden_params, served_from_cache, set_hidden_params +from litellm.litellm_core_utils.litellm_logging import BilledRoundsSpend, UncachedPrice +from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj +from litellm.responses.mcp.parked_stream_dispatch import ( + DISPATCH_RELEASE, + HELD_ROUND_SPEND, + IN_FLIGHT_ROUND, + PARKED_DISPATCH, + SERVED_FROM_CACHE, + DeferredStreamRelease, + parked_success, +) +from litellm.responses.mcp.round_billing import billed_for_every_round, merged_chat_usage, summed_cost_breakdown +from litellm.responses.streaming_iterator import spawn_logging_task +from litellm.types.utils import ModelResponse, ModelResponseStream, Usage + +Parked: TypeAlias = tuple[object, ...] +PartialAnswer: TypeAlias = Callable[[], object] + + +@dataclass(frozen=True, slots=True) +class _AssembledRound: + response: ModelResponse + usage: Usage + cache_hit: bool + parked: Parked + + +@dataclass(frozen=True, slots=True) +class _PricedRound: + response: ModelResponse + usage: Usage + price: UncachedPrice + cache_hit: bool + + @property + def spend(self) -> BilledRoundsSpend: + return BilledRoundsSpend(self.usage, self.price.cost or 0.0, self.price.cost_breakdown) + + +@dataclass(frozen=True, slots=True) +class _HeldUsageChunk: + chunk: ModelResponseStream + usage: Usage + + +@dataclass(frozen=True, slots=True) +class _FirstRound: + priced: _PricedRound + parked: Parked + logged_alone: Parked + banked: LiteLLMLoggingObj | None = None + usage_chunk: _HeldUsageChunk | None = None + + def client_usage_chunk(self, billed: ModelResponse | None) -> ModelResponseStream | None: + if self.usage_chunk is None: + return None + usage: Final = ( + self.priced.usage.model_copy(update=MappingProxyType({"cost": self.priced.price.cost})) + if billed is None + else merged_chat_usage(self.priced.usage, self.usage_chunk.usage).model_copy( + update=MappingProxyType({"cost": _cost(billed)}) + ) + ) + return self.usage_chunk.chunk.model_copy( + update=MappingProxyType({"id": self.priced.response.id, "usage": usage}) + ) + + +def _cost(response: ModelResponse) -> object: + hidden_params: Final = get_hidden_params(response) + return None if hidden_params is None else hidden_params.get("response_cost") + + +def _assembled_round(parked: Parked | None) -> _AssembledRound | None: + match parked: + case (ModelResponse() as response, cache_hit) if isinstance(usage := getattr(response, "usage", None), Usage): + return _AssembledRound(response, usage, cache_hit is True, parked) + case _: + return None + + +def engaged_round_billing(logging_obj: object) -> "ChatStreamRoundBilling | None": + return ChatStreamRoundBilling(logging_obj) if isinstance(logging_obj, LiteLLMLoggingObj) else None + + +class ChatStreamRoundBilling: + def __init__(self, logging_obj: LiteLLMLoggingObj) -> None: + self._logging_obj: Final = logging_obj + self._first: _FirstRound | None = None + self._partial_follow_up: PartialAnswer | None = None + self._reading_loop: asyncio.AbstractEventLoop | None = None + + @contextmanager + def armed_while_reading(self) -> Generator[None]: + armed: Final = getattr(self._logging_obj, DISPATCH_RELEASE, None) is None + if armed: + setattr(self._logging_obj, DISPATCH_RELEASE, self._log_success) + self._reading_loop = asyncio.get_running_loop() + try: + yield + finally: + if armed and self._peek_parked() is None: + self._disarm() + + def complete_first_round(self) -> None: + assembled: Final = _assembled_round(self._peek_parked()) + if assembled is not None: + self._first = _FirstRound(self._priced(assembled), assembled.parked, assembled.parked) + setattr(self._logging_obj, HELD_ROUND_SPEND, self._held_round_one_spend) + + def bank_first_round(self) -> BilledRoundsSpend | None: + self._bank_apart_from_the_follow_up() + return None if self._first is None else self._first.priced.spend + + def client_chunk(self, chunk: ModelResponseStream) -> ModelResponseStream | None: + usage: Final[object] = getattr(chunk, "usage", None) + if self._first is None or not isinstance(usage, Usage): + return chunk + self._first = replace(self._first, usage_chunk=_HeldUsageChunk(chunk, usage)) + return None + + def finish(self) -> ModelResponseStream | None: + first: Final = self._first + billed: Final = self._release_rounds() + return None if first is None else first.client_usage_chunk(billed) + + def fail(self) -> None: + self._bank_apart_from_the_follow_up() + first: Final = self._first + if first is None or first.banked is None or not self._holds_round_one_alone(first): + self.finish() + return + self._first = None + setattr(self._logging_obj, PARKED_DISPATCH, None) + self._log_through_the_release((*first.parked, first.banked.guardrail_scoped_copy())) + + def follow_up_started(self, partial_follow_up: PartialAnswer) -> None: + self._partial_follow_up = partial_follow_up + setattr(self._logging_obj, IN_FLIGHT_ROUND, self.bill_in_flight_round) + + async def bill_in_flight_round(self) -> None: + partial_follow_up: Final = self._partial_follow_up + if partial_follow_up is None or self._first_round_held_alone() is None: + return + final_round: Final = _assembled_round((await asyncify(partial_follow_up)(), False)) + first: Final = self._first_round_held_alone() + if final_round is None or first is None: + return + self._first = None + setattr( + self._logging_obj, PARKED_DISPATCH, self._billed_for_both_rounds(first.priced, self._priced(final_round)) + ) + + def release_abandoned(self) -> None: + if not self._owns_release(): + return + if is_event_loop_running() or self._idle_reading_loop() is not None: + self.finish() + return + setattr(self._logging_obj, PARKED_DISPATCH, None) + self._first = None + self._disarm() + + def _bank_apart_from_the_follow_up(self) -> None: + first: Final = self._first + if first is None or first.banked is not None or self._peek_parked() is not first.parked: + return + banked: Final = self._logging_obj.attempt_scoped_copy() + if not self._owns_release(): + self._first = replace(first, banked=banked) + return + logged_alone: Final = (*first.parked, banked) + self._first = replace(first, logged_alone=logged_alone, banked=banked) + setattr(self._logging_obj, PARKED_DISPATCH, logged_alone) + + def _release_rounds(self) -> ModelResponse | None: + parked: Final = self._take_parked() + first: Final = self._first + self._first = None + if parked is None: + self._disarm() + return None + if first is None or parked is first.logged_alone: + self._release(parked) + return None + final_round: Final = _assembled_round(parked) + if final_round is None: + self._release(first.logged_alone) + return None + billed, cache_hit = self._billed_for_both_rounds(first.priced, self._priced(final_round)) + self._release((billed, cache_hit)) + return billed + + def _billed_for_both_rounds(self, first: _PricedRound, final: _PricedRound) -> tuple[ModelResponse, bool]: + first_breakdown: Final = first.price.cost_breakdown + final_breakdown: Final = final.price.cost_breakdown + rounds: Final = billed_for_every_round(first.response, first_breakdown, final.response, final_breakdown) + usage: Final[object] = getattr(rounds, "usage", None) + charged_usage: Final = ( + usage.model_copy(update=MappingProxyType({"cost": _cost(rounds)})) if isinstance(usage, Usage) else usage + ) + billed: Final = rounds.model_copy(update=MappingProxyType({"id": first.response.id, "usage": charged_usage})) + cache_hit: Final = first.cache_hit and final.cache_hit + set_hidden_params(billed, {**(get_hidden_params(rounds) or {}), "cache_hit": cache_hit}) + self._logging_obj.cost_breakdown = summed_cost_breakdown( + first.response, first_breakdown, final.response, final_breakdown + ) + return billed, cache_hit + + def _priced(self, assembled: _AssembledRound) -> _PricedRound: + served: Final = assembled.cache_hit or served_from_cache(assembled.response) + price: Final = SERVED_FROM_CACHE if served else self._logging_obj.uncached_price(assembled.response) + priced: Final = assembled.response.model_copy() + set_hidden_params( + priced, + {**(get_hidden_params(assembled.response) or {}), "response_cost": price.cost, "cache_hit": served}, + ) + return _PricedRound(priced, assembled.usage, price, served) + + def _peek_parked(self) -> Parked | None: + parked: Final[Parked | None] = getattr(self._logging_obj, PARKED_DISPATCH, None) + return parked if isinstance(parked, tuple) else None + + def _take_parked(self) -> Parked | None: + parked: Final = self._peek_parked() + setattr(self._logging_obj, PARKED_DISPATCH, None) + return parked + + def _first_round_held_alone(self) -> _FirstRound | None: + first: Final = self._first + return first if first is not None and self._holds_round_one_alone(first) else None + + def _held_round_one_spend(self) -> BilledRoundsSpend | None: + first: Final = self._first_round_held_alone() + return None if first is None else first.priced.spend + + def _holds_round_one_alone(self, first: _FirstRound) -> bool: + parked: Final = self._peek_parked() + return parked is first.parked or parked is first.logged_alone + + def _owns_release(self) -> bool: + release: Final[object] = getattr(self._logging_obj, DISPATCH_RELEASE, None) + return release == self._log_success + + def _release(self, parked: Parked) -> None: + if getattr(self._logging_obj, DISPATCH_RELEASE, None) is not None and not self._owns_release(): + setattr(self._logging_obj, PARKED_DISPATCH, parked) + return + self._log_through_the_release(parked) + + def _log_through_the_release(self, parked: Parked) -> None: + armed: Final[DeferredStreamRelease | None] = getattr(self._logging_obj, DISPATCH_RELEASE, None) + setattr(self._logging_obj, DISPATCH_RELEASE, None) + logged: Final = (self._log_success if armed is None else armed)(*parked) + idle_reading_loop: Final = None if is_event_loop_running() else self._idle_reading_loop() + if idle_reading_loop is not None: + idle_reading_loop.run_until_complete(logged) + return + spawn_logging_task(asyncio.get_running_loop(), logged, task_name="Chat completions MCP stream success log") + + def _idle_reading_loop(self) -> asyncio.AbstractEventLoop | None: + loop: Final = self._reading_loop + return loop if loop is not None and not loop.is_closed() and not loop.is_running() else None + + def _disarm(self) -> None: + if self._owns_release(): + setattr(self._logging_obj, DISPATCH_RELEASE, None) + + async def _log_success(self, *parked: object) -> None: + success: Final = parked_success(parked) + if success is not None: + await success.dispatch(self._logging_obj) diff --git a/litellm/responses/mcp/parked_stream_dispatch.py b/litellm/responses/mcp/parked_stream_dispatch.py new file mode 100644 index 00000000000..792363f0bab --- /dev/null +++ b/litellm/responses/mcp/parked_stream_dispatch.py @@ -0,0 +1,72 @@ +from collections.abc import Coroutine +from dataclasses import dataclass +from typing import Final, Protocol + +from litellm.litellm_core_utils.litellm_logging import BilledRoundsSpend, UncachedPrice +from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj + +PARKED_DISPATCH: Final = "_deferred_stream_complete_args" +DISPATCH_RELEASE: Final = "_on_deferred_stream_complete" +IN_FLIGHT_ROUND: Final = "_bill_in_flight_round_on_disconnect" +HELD_ROUND_SPEND: Final = "_held_round_spend_for_failure" +SERVED_FROM_CACHE: Final = UncachedPrice(cost=0.0, cost_breakdown=None) + + +class DispatchesSuccessHandlers(Protocol): + async def dispatch_success_handlers( + self, + result: object = None, + start_time: object = None, + end_time: object = None, + cache_hit: object = None, + prefer_async_handlers: bool = False, + ) -> None: ... + + +class DeferredStreamRelease(Protocol): + def __call__(self, *parked: object) -> Coroutine[object, object, None]: ... + + +class InFlightRoundBiller(Protocol): + def __call__(self) -> Coroutine[object, object, None]: ... + + +class HeldRoundSpend(Protocol): + def __call__(self) -> BilledRoundsSpend | None: ... + + +def success_dispatcher(logger: LiteLLMLoggingObj) -> DispatchesSuccessHandlers: + return logger + + +@dataclass(frozen=True, slots=True) +class ParkedSuccess: + assembled: object + cache_hit: object + logged_on: LiteLLMLoggingObj | None = None + + async def dispatch(self, request_logger: LiteLLMLoggingObj) -> None: + logger: Final = request_logger if self.logged_on is None else self.logged_on + await success_dispatcher(logger).dispatch_success_handlers( + self.assembled, cache_hit=self.cache_hit, start_time=None, end_time=None, prefer_async_handlers=True + ) + + +def in_flight_round_biller(logging_obj: object) -> InFlightRoundBiller | None: + biller: Final[InFlightRoundBiller | None] = getattr(logging_obj, IN_FLIGHT_ROUND, None) + return biller if isinstance(logging_obj, LiteLLMLoggingObj) and callable(biller) else None + + +def held_round_spend(logging_obj: object) -> BilledRoundsSpend | None: + held: Final[HeldRoundSpend | None] = getattr(logging_obj, HELD_ROUND_SPEND, None) + return held() if isinstance(logging_obj, LiteLLMLoggingObj) and callable(held) else None + + +def parked_success(parked: tuple[object, ...]) -> ParkedSuccess | None: + match parked: + case (assembled, cache_hit): + return ParkedSuccess(assembled, cache_hit) + case (assembled, cache_hit, LiteLLMLoggingObj() as logged_on): + return ParkedSuccess(assembled, cache_hit, logged_on) + case _: + return None diff --git a/litellm/responses/mcp/stream_round_billing.py b/litellm/responses/mcp/stream_round_billing.py index 443147230bf..5cb256a8ac0 100644 --- a/litellm/responses/mcp/stream_round_billing.py +++ b/litellm/responses/mcp/stream_round_billing.py @@ -2,11 +2,18 @@ import asyncio from collections.abc import Callable, Coroutine, Sequence from dataclasses import dataclass from datetime import datetime -from typing import Final, Protocol, TypeAlias +from typing import Final, TypeAlias from litellm.litellm_core_utils.hidden_params import get_hidden_params, served_from_cache from litellm.litellm_core_utils.litellm_logging import BilledRoundsDispatch, BilledRoundsSpend, UncachedPrice from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj +from litellm.responses.mcp.parked_stream_dispatch import ( + DISPATCH_RELEASE, + PARKED_DISPATCH, + SERVED_FROM_CACHE, + parked_success, + success_dispatcher, +) from litellm.responses.mcp.round_billing import billed_for_every_round, merged_chat_usage, summed_cost_breakdown from litellm.responses.streaming_iterator import BaseResponsesAPIStreamingIterator, spawn_logging_task from litellm.responses.utils import ResponseAPILoggingUtils @@ -16,25 +23,6 @@ from litellm.types.utils import CostBreakdown, ModelResponse, Usage Parked: TypeAlias = tuple[object, ...] RestoresHeaders: TypeAlias = Callable[[ResponsesAPIStreamingResponse], None] -_PARKED_DISPATCH: Final = "_deferred_stream_complete_args" -_DISPATCH_RELEASE: Final = "_on_deferred_stream_complete" -_SERVED_FROM_CACHE: Final = UncachedPrice(cost=0.0, cost_breakdown=None) - - -class _DispatchesSuccessHandlers(Protocol): - async def dispatch_success_handlers( - self, - result: object = None, - start_time: object = None, - end_time: object = None, - cache_hit: object = None, - prefer_async_handlers: bool = False, - ) -> None: ... - - -def _success_dispatcher(logger: LiteLLMLoggingObj) -> _DispatchesSuccessHandlers: - return logger - def _event_loop_running() -> bool: try: @@ -198,8 +186,8 @@ class StreamedRoundBilling: self._billed: _BilledRounds | None = None self._banked: Parked | None = None self._banked_row: _RoundRow | None = None - if getattr(logging_obj, _DISPATCH_RELEASE, None) is None: - setattr(logging_obj, _DISPATCH_RELEASE, self._run_parked) + if getattr(logging_obj, DISPATCH_RELEASE, None) is None: + setattr(logging_obj, DISPATCH_RELEASE, self._run_parked) def complete_round( self, @@ -286,7 +274,7 @@ class StreamedRoundBilling: def _price(self, response: ResponsesAPIResponse, round_parked: Parked | None) -> UncachedPrice: if served_from_cache(response): - return _SERVED_FROM_CACHE + return SERVED_FROM_CACHE match round_parked: case (ModelResponse() as assembled, bool()): return self._logging_obj.uncached_price(assembled) @@ -297,7 +285,7 @@ class StreamedRoundBilling: self._banked = parked self._banked_row = row if parked is not None: - setattr(self._logging_obj, _PARKED_DISPATCH, parked) + setattr(self._logging_obj, PARKED_DISPATCH, parked) def _bank_apart_from_later_rounds(self, billed: _BilledRounds) -> None: row: Final = self._banked_row @@ -320,7 +308,7 @@ class StreamedRoundBilling: @staticmethod async def _log_success(success: _SuccessLog) -> None: success.carry_bridged_usage() - await _success_dispatcher(success.logger).dispatch_success_handlers( + await success_dispatcher(success.logger).dispatch_success_handlers( success.row.logged, start_time=success.logger.start_time, end_time=success.row.end_time, @@ -329,10 +317,10 @@ class StreamedRoundBilling: ) def _take_parked(self) -> Parked | None: - parked: Final[Parked | None] = getattr(self._logging_obj, _PARKED_DISPATCH, None) + parked: Final[Parked | None] = getattr(self._logging_obj, PARKED_DISPATCH, None) if not isinstance(parked, tuple): return None - setattr(self._logging_obj, _PARKED_DISPATCH, None) + setattr(self._logging_obj, PARKED_DISPATCH, None) return parked async def _run_parked(self, *parked: object) -> None: @@ -341,29 +329,27 @@ class StreamedRoundBilling: await dispatch case (dispatch,) if asyncio.iscoroutine(dispatch): await dispatch - case (assembled, cache_hit): - await _success_dispatcher(self._logging_obj).dispatch_success_handlers( - assembled, cache_hit=cache_hit, start_time=None, end_time=None, prefer_async_handlers=True - ) + case _ if (success := parked_success(parked)) is not None: + await success.dispatch(self._logging_obj) case _: return def _owns_release(self) -> bool: - release: Final[object] = getattr(self._logging_obj, _DISPATCH_RELEASE, None) + release: Final[object] = getattr(self._logging_obj, DISPATCH_RELEASE, None) return release == self._run_parked def _release(self, parked: Parked) -> None: - if getattr(self._logging_obj, _DISPATCH_RELEASE, None) is not None and not self._owns_release(): - setattr(self._logging_obj, _PARKED_DISPATCH, parked) + if getattr(self._logging_obj, DISPATCH_RELEASE, None) is not None and not self._owns_release(): + setattr(self._logging_obj, PARKED_DISPATCH, parked) return - setattr(self._logging_obj, _DISPATCH_RELEASE, None) + setattr(self._logging_obj, DISPATCH_RELEASE, None) spawn_logging_task( asyncio.get_running_loop(), self._run_parked(*parked), task_name="Responses MCP stream success logging" ) def _disarm(self) -> None: if self._owns_release(): - setattr(self._logging_obj, _DISPATCH_RELEASE, None) + setattr(self._logging_obj, DISPATCH_RELEASE, None) @staticmethod def _close(parked: Parked | None) -> None: diff --git a/tests/integration/_support/hook_count_probe.py b/tests/integration/_support/hook_count_probe.py index 0d4056e8fb9..3cd72f1a6ba 100644 --- a/tests/integration/_support/hook_count_probe.py +++ b/tests/integration/_support/hook_count_probe.py @@ -75,6 +75,24 @@ class BlockingPostCallGuardrail(CustomGuardrail): raise HTTPException(status_code=400, detail="blocked by the hook-count guardrail") +class UsageRecordingPostCallGuardrail(CustomGuardrail): + """Records the token usage of the response it is handed in the request's guardrail information, so the spend row + shows which response the guardrail saw.""" + + async def async_post_call_success_hook( + self, data: dict[str, object], user_api_key_dict: object, response: object + ) -> None: + usage: Final = getattr(response, "usage", None) + self.add_standard_logging_guardrail_information_to_request_data( + guardrail_json_response={ + "prompt_tokens": getattr(usage, "prompt_tokens", None), + "completion_tokens": getattr(usage, "completion_tokens", None), + }, + request_data=data, + guardrail_status="success", + ) + + probe: Final = HookCountProbe() sync_only_probe: Final = SyncOnlyHookCountProbe() diff --git a/tests/integration/mcp/test_mcp_llm_endpoints.py b/tests/integration/mcp/test_mcp_llm_endpoints.py index 4b967802893..f6743a8751a 100644 --- a/tests/integration/mcp/test_mcp_llm_endpoints.py +++ b/tests/integration/mcp/test_mcp_llm_endpoints.py @@ -46,9 +46,10 @@ from pydantic import JsonValue, TypeAdapter pytestmark: Final = [pytest.mark.xdist_group("mcp_llm_private_gateway"), pytest.mark.shares_worker_state] -@pytest.fixture(scope="session") -def _llm_gateway(tmp_path_factory: pytest.TempPathFactory) -> Iterator[tuple[Gateway, Mapping[str, str]]]: - directory: Final = tmp_path_factory.mktemp("mcp-llm-gateway") +@contextmanager +def _private_gateway( + directory: Path, guardrails: Sequence[Mapping[str, JsonValue]] = () +) -> Iterator[tuple[Gateway, Mapping[str, str]]]: stock: Final = JSON_OBJECT.validate_python(yaml.safe_load(Path("tests/integration/proxy_config.yaml").read_text())) config: Final = directory / "stored-responses.yaml" (directory / f"{STREAM_BLOCK_MODULE}.py").write_text(STREAM_BLOCK_SOURCE) @@ -68,7 +69,8 @@ def _llm_gateway(tmp_path_factory: pytest.TempPathFactory) -> Iterator[tuple[Gat { "guardrail_name": STREAM_BLOCK, "litellm_params": {"guardrail": f"{STREAM_BLOCK_MODULE}.BlockGuard", "mode": "post_call"}, - } + }, + *guardrails, ], } ) @@ -81,15 +83,61 @@ def _llm_gateway(tmp_path_factory: pytest.TempPathFactory) -> Iterator[tuple[Gat yield private, services -@pytest.fixture -def gateway(_llm_gateway: tuple[Gateway, Mapping[str, str]]) -> Iterator[Gateway]: - private, services = _llm_gateway +@contextmanager +def _serving(served: tuple[Gateway, Mapping[str, str]]) -> Iterator[Gateway]: + private, services = served with pytest.MonkeyPatch.context() as environment: for name, value in services.items(): environment.setenv(name, value) yield private +@pytest.fixture(scope="session") +def _llm_gateway(tmp_path_factory: pytest.TempPathFactory) -> Iterator[tuple[Gateway, Mapping[str, str]]]: + with _private_gateway(tmp_path_factory.mktemp("mcp-llm-gateway")) as served: + yield served + + +@pytest.fixture +def gateway(_llm_gateway: tuple[Gateway, Mapping[str, str]]) -> Iterator[Gateway]: + with _serving(_llm_gateway) as private: + yield private + + +ROUND_USAGE_GUARDRAIL: Final = "mcp-llm-round-usage" +CONTENT_FILTER: Final = f"mcp-llm-content-filter-{uuid.uuid4().hex[:8]}" + + +@pytest.fixture(scope="module") +def _guarded_llm_gateway(tmp_path_factory: pytest.TempPathFactory) -> Iterator[tuple[Gateway, Mapping[str, str]]]: + round_usage: Final[Mapping[str, JsonValue]] = { + "guardrail_name": ROUND_USAGE_GUARDRAIL, + "litellm_params": { + "guardrail": "integration._support.hook_count_probe.UsageRecordingPostCallGuardrail", + "mode": "post_call", + "default_on": True, + }, + } + content_filter: Final[Mapping[str, JsonValue]] = { + "guardrail_name": CONTENT_FILTER, + "litellm_params": { + "guardrail": "litellm_content_filter", + "mode": "post_call", + "default_on": False, + "blocked_words": [{"keyword": "forbidden-keeper", "action": "BLOCK"}], + }, + } + guardrails: Final = (round_usage, content_filter) + with _private_gateway(tmp_path_factory.mktemp("mcp-llm-guarded-gateway"), guardrails) as served: + yield served + + +@pytest.fixture +def guarded_gateway(_guarded_llm_gateway: tuple[Gateway, Mapping[str, str]]) -> Iterator[Gateway]: + with _serving(_guarded_llm_gateway) as private: + yield private + + Surface = Literal["chat", "responses", "messages", "messages_bridge", "responses_bridge"] SURFACES: Final[tuple[Surface, ...]] = ("chat", "responses", "messages", "messages_bridge") ADD: Final = {"a": 2, "b": 3} @@ -1032,6 +1080,184 @@ def test_nonstream_failed_follow_up_bills_round_one_on_the_failure_row( ) +def _streamed_chunks(response: httpx.Response) -> tuple[dict[str, object], ...]: + return tuple( + json.loads(line.removeprefix("data: ")) + for line in response.text.splitlines() + if line.startswith("data: ") and line != "data: [DONE]" + ) + + +def test_streaming_chat_auto_execute_logs_the_answer_and_bills_both_rounds(gateway: Gateway) -> None: + billing: Final = Billing(FIRST_ROUND, FOLLOW_UP_ROUND) + with _rig(gateway, "chat", billing=billing, prices=PRICES) as rig: + response: Final = rig.send(_granted_key(rig), [AUTO], stream=True, stream_options={"include_usage": True}) + assert response.status_code == 200, response.text + chunks: Final = _streamed_chunks(response) + text: Final = "".join( + str(chunk["choices"][0]["delta"].get("content") or "") for chunk in chunks if chunk.get("choices") + ) + assert text == ANSWER, response.text + usages: Final = tuple(chunk["usage"] for chunk in chunks if chunk.get("usage")) + assert usages, response.text + rows: Final = _spend_rows(response.headers["x-litellm-call-id"], "acompletion", "success") + assert len(rows) == 1, rows + row: Final = rows[0] + assert row["request_id"] == chunks[0]["id"], (row, chunks[0]) + assert (row["prompt_tokens"], row["completion_tokens"]) == ( + FIRST_ROUND[0] + FOLLOW_UP_ROUND[0], + FIRST_ROUND[1] + FOLLOW_UP_ROUND[1], + ), row + assert float(str(row["spend"])) == pytest.approx(_price(FIRST_ROUND, FOLLOW_UP_ROUND)), row + final_usage: Final = usages[-1] + assert (final_usage["prompt_tokens"], final_usage["completion_tokens"], final_usage["total_tokens"]) == ( + row["prompt_tokens"], + row["completion_tokens"], + sum((*FIRST_ROUND, *FOLLOW_UP_ROUND)), + ), usages + + +def test_streaming_chat_auto_execute_whose_follow_up_fails_bills_round_one_once(gateway: Gateway) -> None: + billing: Final = Billing(FIRST_ROUND, FOLLOW_UP_ROUND, follow_up_fails=True) + with _rig(gateway, "chat", billing=billing, prices=PRICES) as rig: + response: Final = rig.send(_granted_key(rig), [AUTO], stream=True, stream_options={"include_usage": True}) + assert FOLLOW_UP_OUTAGE in response.text, response.text + call_id: Final = response.headers["x-litellm-call-id"] + succeeded: Final = _spend_rows(call_id, "acompletion", "success") + assert len(succeeded) == 1, succeeded + assert (succeeded[0]["prompt_tokens"], succeeded[0]["completion_tokens"]) == FIRST_ROUND, succeeded + assert float(str(succeeded[0]["spend"])) == pytest.approx(_price(FIRST_ROUND)), succeeded + failed: Final = _spend_rows(call_id, "acompletion", "failure") + assert [float(str(row["spend"])) for row in failed] == [0.0], failed + + +def _guardrail_information(metadata: JsonValue) -> tuple[JsonValue, ...]: + entries: Final = object_value(metadata).get("guardrail_information") + return tuple(entries) if isinstance(entries, list) else () + + +def _round_usage_seen_by_the_guardrail(call_id: str) -> list[JsonValue]: + rows: Final = read_rows( + 'SELECT metadata FROM "LiteLLM_SpendLogs" WHERE litellm_call_id=%s AND call_type=%s AND status=%s', + (call_id, "acompletion", "success"), + ) + entries: Final = tuple(chain.from_iterable(_guardrail_information(row["metadata"]) for row in rows)) + return [ + entry.get("guardrail_response") + for entry in entries + if isinstance(entry, dict) and entry.get("guardrail_name") == ROUND_USAGE_GUARDRAIL + ] + + +def test_streaming_chat_auto_execute_whose_follow_up_fails_runs_the_post_call_guardrail_on_round_one( + guarded_gateway: Gateway, +) -> None: + billing: Final = Billing(FIRST_ROUND, FOLLOW_UP_ROUND, follow_up_fails=True) + with _rig(guarded_gateway, "chat", billing=billing, prices=PRICES) as rig: + response: Final = rig.send(_granted_key(rig), [AUTO], stream=True, stream_options={"include_usage": True}) + assert FOLLOW_UP_OUTAGE in response.text, response.text + call_id: Final = response.headers["x-litellm-call-id"] + succeeded: Final = _spend_rows(call_id, "acompletion", "success") + assert len(succeeded) == 1, succeeded + assert (succeeded[0]["prompt_tokens"], succeeded[0]["completion_tokens"]) == FIRST_ROUND, succeeded + assert _round_usage_seen_by_the_guardrail(call_id) == [ + {"prompt_tokens": FIRST_ROUND[0], "completion_tokens": FIRST_ROUND[1]} + ] + + +def _chat_rows(call_id: str, statuses: Sequence[str]) -> list[dict[str, JsonValue]]: + return eventually( + lambda: read_rows( + 'SELECT status, metadata FROM "LiteLLM_SpendLogs" WHERE litellm_call_id=%s AND call_type=%s', + (call_id, "acompletion"), + ), + lambda found: _statuses(found) == sorted(statuses), + seconds=70, + ) + + +def _entries_of(guardrail_name: str, row: Mapping[str, JsonValue]) -> tuple[dict[str, JsonValue], ...]: + return tuple( + entry + for entry in _guardrail_information(row["metadata"]) + if isinstance(entry, dict) and entry.get("guardrail_name") == guardrail_name + ) + + +def test_streaming_chat_auto_execute_whose_follow_up_fails_counts_its_post_call_content_filter_once( + guarded_gateway: Gateway, +) -> None: + billing: Final = Billing(FIRST_ROUND, FOLLOW_UP_ROUND, follow_up_fails=True) + with _rig(guarded_gateway, "chat", billing=billing, prices=PRICES) as rig: + response: Final = rig.send( + _granted_key(rig), + [AUTO], + stream=True, + stream_options={"include_usage": True}, + guardrails=[CONTENT_FILTER], + ) + assert FOLLOW_UP_OUTAGE in response.text, response.text + rows: Final = _chat_rows(response.headers["x-litellm-call-id"], ("success", "failure")) + filtered: Final = tuple( + (row["status"], entry.get("guardrail_status")) for row in rows for entry in _entries_of(CONTENT_FILTER, row) + ) + assert filtered == (("failure", "guardrail_failed_to_respond"),), rows + (guardrail_id,) = frozenset( + str(entry.get("guardrail_id") or CONTENT_FILTER) + for row in rows + for entry in _entries_of(CONTENT_FILTER, row) + ) + counted: Final = eventually( + lambda: read_rows( + 'SELECT requests_evaluated, flagged_count FROM "LiteLLM_DailyGuardrailMetrics" WHERE guardrail_id=%s', + (guardrail_id,), + ), + lambda found: bool(found), + seconds=70, + ) + assert counted == [{"requests_evaluated": 1, "flagged_count": 1}], counted + + +@pytest.mark.parametrize("include_usage", (True, False), ids=("include-usage", "no-usage")) +def test_streaming_chat_auto_execute_answered_in_one_round_ends_with_the_usage_it_billed( + gateway: Gateway, include_usage: bool +) -> None: + billing: Final = Billing(FIRST_ROUND, FOLLOW_UP_ROUND) + with _rig(gateway, "chat", billing=billing, prices=PRICES) as rig: + response: Final = rig.send( + _granted_key(rig), + [AUTO], + prompt=f"{DIRECT} {rig.alias}", + stream=True, + stream_options={"include_usage": include_usage}, + ) + assert response.status_code == 200, response.text + assert response.text.rstrip().endswith("data: [DONE]"), response.text + chunks: Final = _streamed_chunks(response) + text: Final = "".join( + str(chunk["choices"][0]["delta"].get("content") or "") for chunk in chunks if chunk.get("choices") + ) + assert text == ANSWER, response.text + rows: Final = _spend_rows(response.headers["x-litellm-call-id"], "acompletion", "success") + assert len(rows) == 1 and (rows[0]["prompt_tokens"], rows[0]["completion_tokens"]) == FIRST_ROUND, rows + usage_at: Final = tuple(index for index, chunk in enumerate(chunks) if chunk.get("usage")) + if not include_usage: + assert usage_at == (), response.text + return + assert usage_at == (len(chunks) - 1,), response.text + finished_at: Final = tuple( + index + for index, chunk in enumerate(chunks) + if chunk.get("choices") and chunk["choices"][0].get("finish_reason") is not None + ) + assert finished_at and finished_at[-1] < usage_at[0], response.text + usage: Final = chunks[-1]["usage"] + assert (usage["prompt_tokens"], usage["completion_tokens"], usage["total_tokens"]) == ( + *FIRST_ROUND, + sum(FIRST_ROUND), + ), usage + + NO_CACHE: Final = {"no-cache": True} @@ -1269,6 +1495,27 @@ def test_streaming_responses_follow_up_broken_mid_stream_bills_round_one_once_an assert float(str(failure["spend"])) == pytest.approx(_price(failed_round)), rows +def test_streaming_chat_follow_up_broken_mid_answer_bills_round_one_and_its_partial_answer_with_one_fixed_margin( + gateway: Gateway, +) -> None: + billing: Final = Billing(FIRST_ROUND, FOLLOW_UP_ROUND, follow_up_breaks=True) + with _rig(gateway, "chat", upstream="messages_bridge", billing=billing, prices=PRICES) as rig: + response: Final = rig.send(_granted_key(rig), [AUTO], stream=True, stream_options={"include_usage": True}) + assert "the sum" in response.text, response.text + assert ANSWER not in response.text, response.text + assert len(rig.upstream_tools()) == 2 + round_one: Final = _price(FIRST_ROUND) + FIXED_MARGIN + rows: Final = _model_rows(response.headers["x-litellm-call-id"], _logs_a_failure_billing(round_one)) + assert _statuses(rows) == ["failure", "success"], rows + (success,) = (row for row in rows if row["status"] == "success") + (failure,) = (row for row in rows if row["status"] == "failure") + assert (success["prompt_tokens"], success["completion_tokens"]) == FIRST_ROUND, rows + assert float(str(success["spend"])) == pytest.approx(round_one), rows + partial_answer: Final = (int(str(failure["prompt_tokens"])), int(str(failure["completion_tokens"]))) + assert min(partial_answer) > 0, rows + assert float(str(failure["spend"])) == pytest.approx(_price(partial_answer)), rows + + @pytest.mark.parametrize("upstream", STREAMED_UPSTREAMS, ids=STREAMED_IDS) def test_streaming_responses_blocked_by_a_post_call_guardrail_bills_every_round_once( gateway: Gateway, upstream: StreamedUpstream diff --git a/tests/unit/proxy/test_common_request_processing.py b/tests/unit/proxy/test_common_request_processing.py index 2d08609c93b..2db8941d304 100644 --- a/tests/unit/proxy/test_common_request_processing.py +++ b/tests/unit/proxy/test_common_request_processing.py @@ -83,6 +83,7 @@ from litellm.proxy.hooks.model_max_budget_limiter import ( team_member_budget_entity_id, ) from litellm.proxy.utils import ProxyLogging +from litellm.responses.mcp.parked_stream_dispatch import IN_FLIGHT_ROUND from litellm.router import Router from litellm.litellm_core_utils.streaming_handler import CustomStreamWrapper from litellm.utils import ModelResponseListIterator @@ -9051,6 +9052,33 @@ class TestStreamingClientDisconnectBilling: proxy_logging_obj.arelease_max_parallel_requests_on_disconnect.assert_awaited_once() + @pytest.mark.asyncio + async def test_disconnect_whose_in_flight_round_billing_raises_still_closes_the_stream_and_frees_the_slot(self): + import types + + response = await self._start_partial_stream() + closed: Final = AsyncMock() + + async def cannot_price_the_in_flight_round() -> None: + raise RuntimeError("the in-flight round could not be priced") + + setattr(response.logging_obj, IN_FLIGHT_ROUND, cannot_price_the_in_flight_round) + stream: Final = types.SimpleNamespace(chunks=[], messages=None, aclose=closed) + proxy_logging_obj: Final = types.SimpleNamespace(arelease_max_parallel_requests_on_disconnect=AsyncMock()) + + await ProxyBaseLLMRequestProcessing._finalize_streaming_generator_cleanup( + request=None, + request_data={"litellm_logging_obj": response.logging_obj}, + response=stream, + stream_completed=False, + client_disconnected=True, + user_api_key_dict=MagicMock(), + proxy_logging_obj=proxy_logging_obj, + ) + + closed.assert_awaited_once() + proxy_logging_obj.arelease_max_parallel_requests_on_disconnect.assert_awaited_once() + async def _bill_and_collect_success_event(self, prepare=None, request_data=None): recorder = _RecordingSuccessLogger() original_callbacks = litellm.callbacks diff --git a/tests/unit/responses/mcp/test_chat_completions_handler.py b/tests/unit/responses/mcp/test_chat_completions_handler.py index ec5968c2a1a..c0d32742e8a 100644 --- a/tests/unit/responses/mcp/test_chat_completions_handler.py +++ b/tests/unit/responses/mcp/test_chat_completions_handler.py @@ -1,17 +1,25 @@ import asyncio +import gc import json +import queue import sys +import threading import types -from collections.abc import Mapping, Sequence +import warnings +from collections.abc import AsyncIterator, Callable, Iterator, Mapping, Sequence +from contextlib import contextmanager from datetime import datetime from types import MappingProxyType from typing import Final, cast -from unittest.mock import AsyncMock, patch +from unittest.mock import AsyncMock, MagicMock, patch +import httpx import pytest import respx from httpx import Response from mcp.types import Tool as MCPTool +from starlette.responses import StreamingResponse +from starlette.types import Message import litellm from litellm.integrations.custom_logger import CustomLogger @@ -25,7 +33,8 @@ from litellm.responses.mcp.litellm_proxy_mcp_handler import ( LiteLLM_Proxy_MCP_Handler, ) from litellm.responses.utils import ResponsesAPIRequestUtils -from litellm.types.utils import ModelResponse, StandardLoggingPayload +from litellm.types.utils import ModelResponse, ModelResponseStream, StandardLoggingPayload, Usage +from litellm.utils import CustomStreamWrapper @pytest.mark.asyncio @@ -1556,6 +1565,7 @@ _CHAT_ROUND_TWO: Final = MappingProxyType( ), } ) +_NEVER_LOGGED_SECONDS: Final = 60 _CONTEXT_WINDOW_ERROR: Final = MappingProxyType( { "error": MappingProxyType( @@ -1575,16 +1585,31 @@ class _ChatRoundRecorder(CustomLogger): super().__init__() self.logged: tuple[StandardLoggingPayload, ...] = () self.failed: tuple[StandardLoggingPayload, ...] = () + self._recorded: Final = asyncio.Condition() async def async_log_success_event( self, kwargs: Mapping[str, object], response_obj: object, start_time: datetime, end_time: datetime ) -> None: self.logged = (*self.logged, cast(StandardLoggingPayload, kwargs["standard_logging_object"])) + await self._announce() async def async_log_failure_event( self, kwargs: Mapping[str, object], response_obj: object, start_time: datetime, end_time: datetime ) -> None: self.failed = (*self.failed, cast(StandardLoggingPayload, kwargs["standard_logging_object"])) + await self._announce() + + async def _announce(self) -> None: + async with self._recorded: + self._recorded.notify_all() + + async def settled(self, logged: int = 0, failed: int = 0) -> None: + async with self._recorded: + await asyncio.wait_for( + self._recorded.wait_for(lambda: len(self.logged) >= logged and len(self.failed) >= failed), + timeout=_NEVER_LOGGED_SECONDS, + ) + await _settled() def _chat_round_cost(body: Mapping[str, object]) -> float: @@ -1600,18 +1625,20 @@ async def _settled() -> None: await asyncio.wait_for(GLOBAL_LOGGING_WORKER.flush(), timeout=10) -async def _auto_execute_chat_echo( - respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch, rounds: Sequence[Response], call_id: str -) -> tuple[_ChatRoundRecorder, ModelResponse | Exception]: +def _serve_chat_rounds( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch, rounds: Sequence[Response] +) -> None: monkeypatch.setattr(litellm, "disable_aiohttp_transport", True) monkeypatch.setenv("DISABLE_AIOHTTP_TRANSPORT", "True") litellm.in_memory_llm_clients_cache.flush_cache() respx_mock.post("https://mcp-chat-rounds.example.com/v1/chat/completions").mock(side_effect=list(rounds)) - recorder: Final = _ChatRoundRecorder() - logging_obj: Final = Logging( + + +def _chat_rounds_logging_obj(recorder: _ChatRoundRecorder, call_id: str, stream: bool) -> Logging: + return Logging( model="openai/gpt-5.6", messages=[{"role": "user", "content": "Call echo with PING"}], - stream=False, + stream=stream, call_type="acompletion", start_time=datetime(2026, 1, 1), litellm_call_id=call_id, @@ -1619,6 +1646,10 @@ async def _auto_execute_chat_echo( dynamic_async_success_callbacks=[recorder], dynamic_async_failure_callbacks=[recorder], ) + + +@contextmanager +def _echo_tool_served() -> Iterator[None]: echo_tool: Final = MCPTool.model_validate( {"name": "echo", "inputSchema": {"type": "object", "properties": {"value": {"type": "string"}}}}, by_name=False, @@ -1635,6 +1666,16 @@ async def _auto_execute_chat_echo( AsyncMock(return_value=[{"tool_call_id": "call_echo", "result": "PING", "name": "echo"}]), ), ): + yield + + +async def _auto_execute_chat_echo( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch, rounds: Sequence[Response], call_id: str +) -> tuple[_ChatRoundRecorder, ModelResponse | Exception]: + _serve_chat_rounds(respx_mock, monkeypatch, rounds) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, call_id, stream=False) + with _echo_tool_served(): try: response: Final = await litellm.acompletion( model="openai/gpt-5.6", @@ -1712,3 +1753,994 @@ async def test_non_stream_auto_execute_bills_the_paid_first_round_when_the_follo assert (recorder.failed[0]["prompt_tokens"], recorder.failed[0]["completion_tokens"]) == (71, 19) assert recorder.failed[0]["response_cost"] == pytest.approx(_chat_round_cost(_CHAT_ROUND_ONE)) assert recorder.failed[0]["response_cost"] > 0 + + +_FIXED_MARGIN: Final = 0.001 + + +def _chat_stream_events(body: Mapping[str, object]) -> tuple[Mapping[str, object], ...]: + completed: Final = json.loads(json.dumps(body, default=dict)) + choice: Final = completed["choices"][0] + message: Final = choice["message"] + calls: Final = message.get("tool_calls") or [] + delta: Final = ( + {**message, "tool_calls": [{**call, "index": index} for index, call in enumerate(calls)]} if calls else message + ) + head: Final = {"id": completed["id"], "object": "chat.completion.chunk", "created": 1, "model": completed["model"]} + return ( + {**head, "choices": [{"index": 0, "delta": delta, "finish_reason": None}]}, + {**head, "choices": [{"index": 0, "delta": {}, "finish_reason": choice["finish_reason"]}]}, + {**head, "choices": [], "usage": completed["usage"]}, + ) + + +def _sse(events: Sequence[Mapping[str, object]]) -> bytes: + return b"".join(f"data: {json.dumps(event)}\n\n".encode() for event in events) + + +def _streamed(body: Mapping[str, object]) -> Response: + return Response( + 200, + content=_sse(_chat_stream_events(body)) + b"data: [DONE]\n\n", + headers={"content-type": "text/event-stream"}, + ) + + +class _CutOnceTheAnswerStarts(httpx.AsyncByteStream): + async def __aiter__(self) -> AsyncIterator[bytes]: + yield _sse(_chat_stream_events(_CHAT_ROUND_TWO)[:1]) + raise httpx.ReadError("connection reset mid-stream") + + +def _cut_once_the_answer_starts() -> Response: + return Response(200, stream=_CutOnceTheAnswerStarts(), headers={"content-type": "text/event-stream"}) + + +async def _open_chat_echo_stream( + logging_obj: Logging, metadata: dict[str, object] | None = None +) -> CustomStreamWrapper: + stream: Final = await litellm.acompletion( + model="openai/gpt-5.6", + messages=[{"role": "user", "content": "Call echo with PING"}], + tools=[{"type": "mcp", "server_url": "litellm_proxy/mcp/demo", "require_approval": "never"}], + api_key="sk-mcp-chat-rounds", + api_base="https://mcp-chat-rounds.example.com/v1", + stream=True, + stream_options={"include_usage": True}, + litellm_logging_obj=logging_obj, + metadata=metadata, + ) + assert isinstance(stream, CustomStreamWrapper) + return stream + + +async def _read_all(stream: CustomStreamWrapper) -> tuple[ModelResponseStream, ...]: + return tuple([chunk async for chunk in stream]) + + +async def _read_until(stream: CustomStreamWrapper, content: str) -> None: + async for chunk in stream: + if chunk.choices and chunk.choices[0].delta.content == content: + return + pytest.fail(f"the stream never sent {content!r}") + + +def _streamed_usage(chunks: Sequence[ModelResponseStream]) -> Usage: + usages: Final = tuple(usage for chunk in chunks if isinstance(usage := getattr(chunk, "usage", None), Usage)) + assert len(usages) == 1, usages + return usages[0] + + +def _assert_one_stream_row_bills_both_rounds_once( + logged: Sequence[StandardLoggingPayload], chunks: Sequence[ModelResponseStream] +) -> None: + charged_once: Final = _chat_round_cost(_CHAT_ROUND_ONE) + _chat_round_cost(_CHAT_ROUND_TWO) - _FIXED_MARGIN + assert len(logged) == 1 + row: Final = logged[0] + assert row["id"] == chunks[0].id + assert (row["prompt_tokens"], row["completion_tokens"]) == (71 + 149, 19 + 5) + assert row["metadata"]["usage_object"]["prompt_tokens_details"]["cached_tokens"] == 32 + 64 + assert row["response_cost"] == pytest.approx(charged_once) + assert row["cost_breakdown"]["total_cost"] == pytest.approx(charged_once) + assert row["metadata"]["usage_object"]["cost"] == pytest.approx(row["response_cost"]) + usage: Final = _streamed_usage(chunks) + assert (usage.prompt_tokens, usage.completion_tokens, usage.total_tokens) == (71 + 149, 19 + 5, 90 + 154) + assert usage.cost == pytest.approx(row["response_cost"]) + + +def _prompt_tokens_the_follow_up_sent(respx_mock: respx.MockRouter) -> int: + follow_up_request: Final = json.loads(respx_mock.calls[1].request.content) + return litellm.token_counter(model="gpt-5.6", messages=follow_up_request["messages"]) + + +def _assert_one_row_bills_round_one(logged: Sequence[StandardLoggingPayload]) -> None: + assert len(logged) == 1 + assert (logged[0]["prompt_tokens"], logged[0]["completion_tokens"]) == (71, 19) + assert logged[0]["response_cost"] == pytest.approx(_chat_round_cost(_CHAT_ROUND_ONE)) + + +@pytest.mark.asyncio +async def test_stream_auto_execute_read_to_the_end_logs_one_row_whose_usage_cost_counts_the_fixed_margin_once( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch +): + monkeypatch.setattr(litellm, "cost_margin_config", {"openai": {"fixed_amount": _FIXED_MARGIN}}) + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _streamed(_CHAT_ROUND_TWO))) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-read-to-the-end", stream=True) + + with _echo_tool_served(): + chunks: Final = await _read_all(await _open_chat_echo_stream(logging_obj)) + await recorder.settled(logged=1) + + _assert_one_stream_row_bills_both_rounds_once(recorder.logged, chunks) + + +@pytest.mark.asyncio +async def test_stream_auto_execute_deferred_for_a_post_call_guardrail_logs_both_rounds_once_when_the_proxy_releases( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch +): + from litellm.proxy.common_request_processing import ProxyBaseLLMRequestProcessing + from litellm.proxy.utils import ProxyLogging + + monkeypatch.setattr(litellm, "cost_margin_config", {"openai": {"fixed_amount": _FIXED_MARGIN}}) + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _streamed(_CHAT_ROUND_TWO))) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-rounds-deferred", stream=True) + + with _echo_tool_served(): + stream: Final = await _open_chat_echo_stream(logging_obj) + ProxyBaseLLMRequestProcessing(data={})._arm_deferred_stream_dispatch( + response=stream, route_type="acompletion", user_api_key_dict=MagicMock(), logging_obj=logging_obj + ) + chunks: Final = await _read_all(stream) + await _settled() + logged_before_release: Final = recorder.logged + ProxyLogging.fire_deferred_stream_logging({"litellm_logging_obj": logging_obj}) + await recorder.settled(logged=1) + + assert logged_before_release == () + _assert_one_stream_row_bills_both_rounds_once(recorder.logged, chunks) + + +@pytest.mark.asyncio +async def test_stream_auto_execute_blocked_by_a_post_call_guardrail_bills_both_rounds_on_the_failure_row( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch +): + from litellm.exceptions import GuardrailRaisedException + from litellm.proxy.common_request_processing import ProxyBaseLLMRequestProcessing + from litellm.proxy.utils import ProxyLogging + + monkeypatch.setattr(litellm, "cost_margin_config", {"openai": {"fixed_amount": _FIXED_MARGIN}}) + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _streamed(_CHAT_ROUND_TWO))) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-rounds-blocked", stream=True) + blocked: Final = GuardrailRaisedException(guardrail_name="output-policy", message="blocked") + + with _echo_tool_served(): + stream: Final = await _open_chat_echo_stream(logging_obj) + ProxyBaseLLMRequestProcessing(data={})._arm_deferred_stream_dispatch( + response=stream, route_type="acompletion", user_api_key_dict=MagicMock(), logging_obj=logging_obj + ) + await _read_all(stream) + assert ProxyLogging._discard_deferred_stream_logging_for_failure({"litellm_logging_obj": logging_obj}, blocked) + await logging_obj.async_failure_handler(blocked, "blocked by output-policy") + await recorder.settled(failed=1) + + charged_once: Final = _chat_round_cost(_CHAT_ROUND_ONE) + _chat_round_cost(_CHAT_ROUND_TWO) - _FIXED_MARGIN + assert recorder.logged == () + assert len(recorder.failed) == 1 + assert (recorder.failed[0]["prompt_tokens"], recorder.failed[0]["completion_tokens"]) == (71 + 149, 19 + 5) + assert recorder.failed[0]["response_cost"] == pytest.approx(charged_once) + + +@pytest.mark.asyncio +async def test_stream_auto_execute_rejected_by_the_proxy_during_the_follow_up_bills_round_one_with_its_own_breakdown( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch +): + from fastapi import HTTPException + from litellm.proxy.common_request_processing import ProxyBaseLLMRequestProcessing + from litellm.proxy.utils import ProxyLogging + + monkeypatch.setattr(litellm, "cost_margin_config", {"openai": {"fixed_amount": _FIXED_MARGIN}}) + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _streamed(_CHAT_ROUND_TWO))) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-rejected-mid-follow-up", stream=True) + rejected: Final = HTTPException(status_code=400, detail="rejected by a stream hook") + + with _echo_tool_served(): + stream: Final = await _open_chat_echo_stream(logging_obj) + ProxyBaseLLMRequestProcessing(data={})._arm_deferred_stream_dispatch( + response=stream, route_type="acompletion", user_api_key_dict=MagicMock(), logging_obj=logging_obj + ) + await _read_until(stream, "PING") + assert ProxyLogging._discard_deferred_stream_logging_for_failure({"litellm_logging_obj": logging_obj}, rejected) + await logging_obj.async_failure_handler(rejected, "rejected by a stream hook") + await recorder.settled(failed=1) + + assert recorder.logged == () + assert len(recorder.failed) == 1 + failed: Final = recorder.failed[0] + breakdown: Final = failed["cost_breakdown"] + assert (failed["prompt_tokens"], failed["completion_tokens"]) == (71, 19) + assert failed["response_cost"] == pytest.approx(_chat_round_cost(_CHAT_ROUND_ONE)) + assert breakdown is not None + assert breakdown["total_cost"] == pytest.approx(failed["response_cost"]) + + +@pytest.mark.asyncio +async def test_stream_auto_execute_answered_without_a_tool_call_logs_its_one_round_once_the_answer_finishes( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch +): + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_TWO),)) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-one-round", stream=True) + + with _echo_tool_served(): + stream: Final = await _open_chat_echo_stream(logging_obj) + async for chunk in stream: + if chunk.choices and chunk.choices[0].finish_reason is not None: + break + await recorder.settled(logged=1) + logged_once_the_answer_finished: Final = recorder.logged + after_the_answer: Final = await _read_all(stream) + await _settled() + + assert len(logged_once_the_answer_finished) == 1 + assert recorder.logged == logged_once_the_answer_finished + assert recorder.logged[0]["id"] == _CHAT_ROUND_TWO["id"] + assert (recorder.logged[0]["prompt_tokens"], recorder.logged[0]["completion_tokens"]) == (149, 5) + assert after_the_answer[-1:] and isinstance(after_the_answer[-1].usage, Usage) + usage: Final = _streamed_usage(after_the_answer) + assert (usage.prompt_tokens, usage.completion_tokens, usage.total_tokens) == ( + recorder.logged[0]["prompt_tokens"], + recorder.logged[0]["completion_tokens"], + recorder.logged[0]["total_tokens"], + ) + assert recorder.logged[0]["response_cost"] == pytest.approx(_chat_round_cost(_CHAT_ROUND_TWO)) + assert recorder.failed == () + + +@pytest.mark.asyncio +@pytest.mark.parametrize("closes", (True, False), ids=("aclose", "dropped")) +async def test_stream_auto_execute_bills_the_finished_first_round_when_the_caller_stops_reading_the_follow_up( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch, closes: bool +): + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _streamed(_CHAT_ROUND_TWO))) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-caller-stops", stream=True) + + async def stop_reading_once_the_follow_up_answers() -> None: + stream: Final = await _open_chat_echo_stream(logging_obj) + await _read_until(stream, "PING") + if closes: + await stream.aclose() + + with _echo_tool_served(): + await stop_reading_once_the_follow_up_answers() + await recorder.settled(logged=1) + + _assert_one_row_bills_round_one(recorder.logged) + + +@pytest.mark.asyncio +async def test_stream_auto_execute_whose_first_round_breaks_mid_answer_bills_only_its_failure_row( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch +): + _serve_chat_rounds(respx_mock, monkeypatch, (_cut_once_the_answer_starts(),)) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-first-round-breaks", stream=True) + + with _echo_tool_served(): + stream: Final = await _open_chat_echo_stream(logging_obj) + with pytest.raises(litellm.ServiceUnavailableError): + await _read_all(stream) + await recorder.settled(failed=1) + + assert recorder.logged == () + assert len(recorder.failed) == 1 + assert recorder.failed[0]["completion_tokens"] > 0 + + +@pytest.mark.asyncio +@pytest.mark.parametrize("deferred_for_a_post_call_guardrail", [False, True]) +async def test_stream_auto_execute_bills_a_broken_follow_up_on_its_own_failure_row_with_no_second_fixed_margin( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch, deferred_for_a_post_call_guardrail: bool +): + from litellm.proxy.common_request_processing import ProxyBaseLLMRequestProcessing + from litellm.proxy.utils import ProxyLogging, _failure_fields_to_lift + + monkeypatch.setattr(litellm, "cost_margin_config", {"openai": {"fixed_amount": _FIXED_MARGIN}}) + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _cut_once_the_answer_starts())) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-follow-up-breaks", stream=True) + request_data: Final = MappingProxyType({"litellm_logging_obj": logging_obj}) + + with _echo_tool_served(): + stream: Final = await _open_chat_echo_stream(logging_obj) + if deferred_for_a_post_call_guardrail: + ProxyBaseLLMRequestProcessing(data={})._arm_deferred_stream_dispatch( + response=stream, route_type="acompletion", user_api_key_dict=MagicMock(), logging_obj=logging_obj + ) + with pytest.raises(litellm.ServiceUnavailableError) as broken: + await _read_all(stream) + if not ProxyLogging._discard_deferred_stream_logging_for_failure(dict(request_data), broken.value): + ProxyLogging.fire_deferred_stream_logging(dict(request_data)) + await recorder.settled(logged=1, failed=1) + + _assert_one_row_bills_round_one(recorder.logged) + assert len(recorder.failed) == 1 + failed: Final = recorder.failed[0] + unmargined: Final = ( + litellm.completion_cost( + completion_response=ModelResponse( + model="gpt-5.6", + usage=Usage( + prompt_tokens=failed["prompt_tokens"], + completion_tokens=failed["completion_tokens"], + total_tokens=failed["total_tokens"], + ), + ), + model="gpt-5.6", + custom_llm_provider="openai", + ) + - _FIXED_MARGIN + ) + assert failed["prompt_tokens"] == _prompt_tokens_the_follow_up_sent(respx_mock) + assert failed["completion_tokens"] > 0 + assert failed["response_cost"] == pytest.approx(unmargined) + request_failure: Final = _failure_fields_to_lift(request_data) + assert cast(StandardLoggingPayload, request_failure["standard_logging_object"])["id"] != recorder.logged[0]["id"] + assert request_failure["response_cost"] == pytest.approx(unmargined) + + +@pytest.mark.asyncio +@pytest.mark.parametrize("deferred_for_a_post_call_guardrail", [False, True]) +async def test_stream_auto_execute_whose_follow_up_breaks_reports_the_request_guardrail_result_on_the_failure_row_only( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch, deferred_for_a_post_call_guardrail: bool +): + from litellm.integrations.custom_guardrail import CustomGuardrail + from litellm.proxy.common_request_processing import ProxyBaseLLMRequestProcessing + from litellm.proxy.utils import ProxyLogging + + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _cut_once_the_answer_starts())) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-follow-up-breaks-guarded", stream=True) + metadata: Final[dict[str, object]] = {} + request_data: Final = {"metadata": metadata, "litellm_logging_obj": logging_obj} + content_filter: Final = CustomGuardrail(guardrail_name="request-content-filter", event_hook="post_call") + + with _echo_tool_served(): + stream: Final = await _open_chat_echo_stream(logging_obj, metadata) + if deferred_for_a_post_call_guardrail: + ProxyBaseLLMRequestProcessing(data={})._arm_deferred_stream_dispatch( + response=stream, route_type="acompletion", user_api_key_dict=MagicMock(), logging_obj=logging_obj + ) + with pytest.raises(litellm.ServiceUnavailableError) as broken: + await _read_all(stream) + content_filter.add_standard_logging_guardrail_information_to_request_data( + guardrail_json_response=str(broken.value), + request_data=request_data, + guardrail_status="guardrail_failed_to_respond", + ) + if not ProxyLogging._discard_deferred_stream_logging_for_failure(request_data, broken.value): + ProxyLogging.fire_deferred_stream_logging(request_data) + await recorder.settled(logged=1, failed=1) + + _assert_one_row_bills_round_one(recorder.logged) + assert recorder.logged[0]["guardrail_information"] is None + assert [entry["guardrail_name"] for entry in recorder.failed[0]["guardrail_information"] or ()] == [ + "request-content-filter" + ] + + +@pytest.mark.asyncio +async def test_stream_auto_execute_served_through_the_proxy_stream_hook_logs_both_rounds_once_when_it_ends( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch +): + from litellm.caching.caching import DualCache + from litellm.proxy._types import UserAPIKeyAuth + from litellm.proxy.utils import ProxyLogging + + monkeypatch.setattr(litellm, "cost_margin_config", {"openai": {"fixed_amount": _FIXED_MARGIN}}) + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _streamed(_CHAT_ROUND_TWO))) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-proxy-stream-hook", stream=True) + + with _echo_tool_served(): + served: Final = ProxyLogging(user_api_key_cache=DualCache()).async_post_call_streaming_iterator_hook( + response=await _open_chat_echo_stream(logging_obj), + user_api_key_dict=UserAPIKeyAuth(), + request_data={"litellm_logging_obj": logging_obj}, + ) + chunks: Final = tuple([chunk async for chunk in served]) + await recorder.settled(logged=1) + + _assert_one_stream_row_bills_both_rounds_once(recorder.logged, chunks) + + +@pytest.mark.asyncio +async def test_stream_auto_execute_whose_follow_up_streams_nothing_still_bills_the_first_round_once( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch +): + nothing_streamed: Final = Response(200, content=b"data: [DONE]\n\n", headers={"content-type": "text/event-stream"}) + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), nothing_streamed)) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-empty-follow-up", stream=True) + + with _echo_tool_served(): + await _read_all(await _open_chat_echo_stream(logging_obj)) + await recorder.settled(logged=1) + + _assert_one_row_bills_round_one(recorder.logged) + + +_ANSWER_WITH_UNREADABLE_AUDIO: Final = MappingProxyType( + { + **_CHAT_ROUND_TWO, + "choices": ( + MappingProxyType( + { + "index": 0, + "finish_reason": "stop", + "message": MappingProxyType( + { + "role": "assistant", + "content": "PING", + "audio": MappingProxyType({"id": "audio_echo", "data": "not base64", "transcript": "PING"}), + } + ), + } + ), + ), + } +) + + +@pytest.mark.asyncio +async def test_stream_auto_execute_whose_follow_up_cannot_be_assembled_ends_with_the_usage_of_the_round_it_billed( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch +): + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _streamed(_ANSWER_WITH_UNREADABLE_AUDIO))) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-follow-up-unassembled", stream=True) + + with _echo_tool_served(): + chunks: Final = await _read_all(await _open_chat_echo_stream(logging_obj)) + await recorder.settled(logged=1) + + _assert_one_row_bills_round_one(recorder.logged) + assert any(chunk.choices and chunk.choices[0].delta.content == "PING" for chunk in chunks) + usage: Final = _streamed_usage(chunks) + assert (usage.prompt_tokens, usage.completion_tokens, usage.total_tokens) == (71, 19, 90) + assert usage.cost == pytest.approx(recorder.logged[0]["response_cost"]) + + +@pytest.mark.asyncio +@pytest.mark.parametrize("deferred_for_a_post_call_guardrail", [False, True]) +@pytest.mark.parametrize("answer", [_CHAT_ROUND_TWO, _ANSWER_WITH_UNREADABLE_AUDIO], ids=("assembled", "unassembled")) +async def test_stream_auto_execute_sends_its_usage_chunk_under_the_id_of_the_row_it_billed( + respx_mock: respx.MockRouter, + monkeypatch: pytest.MonkeyPatch, + deferred_for_a_post_call_guardrail: bool, + answer: Mapping[str, object], +): + from litellm.proxy.common_request_processing import ProxyBaseLLMRequestProcessing + from litellm.proxy.utils import ProxyLogging + + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _streamed(answer))) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-usage-chunk-id", stream=True) + + with _echo_tool_served(): + stream: Final = await _open_chat_echo_stream(logging_obj) + if deferred_for_a_post_call_guardrail: + ProxyBaseLLMRequestProcessing(data={})._arm_deferred_stream_dispatch( + response=stream, route_type="acompletion", user_api_key_dict=MagicMock(), logging_obj=logging_obj + ) + chunks: Final = await _read_all(stream) + ProxyLogging.fire_deferred_stream_logging({"litellm_logging_obj": logging_obj}) + await recorder.settled(logged=1) + + usage_chunks: Final = tuple(chunk for chunk in chunks if isinstance(getattr(chunk, "usage", None), Usage)) + answer_chunks: Final = tuple(chunk for chunk in chunks if chunk.choices and chunk.choices[0].delta.content == "PING") + assert len(recorder.logged) == 1 + assert [chunk.id for chunk in answer_chunks] == [_CHAT_ROUND_TWO["id"]] + assert len(usage_chunks) == 1 + assert usage_chunks[0].id == chunks[0].id == recorder.logged[0]["id"] == _CHAT_ROUND_ONE["id"] + + +def _assert_one_row_bills_round_one_and_the_partial_answer( + recorder: _ChatRoundRecorder, respx_mock: respx.MockRouter +) -> None: + assert len(recorder.logged) == 1 + row: Final = recorder.logged[0] + assert row["id"] == _CHAT_ROUND_ONE["id"] + partial_prompt, partial_completion = (row["prompt_tokens"] - 71, row["completion_tokens"] - 19) + assert partial_prompt == _prompt_tokens_the_follow_up_sent(respx_mock) + assert partial_completion > 0 + partial_answer: Final = litellm.completion_cost( + completion_response=ModelResponse( + model="gpt-5.6", + usage=Usage( + prompt_tokens=partial_prompt, + completion_tokens=partial_completion, + total_tokens=partial_prompt + partial_completion, + ), + ), + model="gpt-5.6", + custom_llm_provider="openai", + ) + assert row["response_cost"] == pytest.approx(_chat_round_cost(_CHAT_ROUND_ONE) + partial_answer - _FIXED_MARGIN) + assert recorder.failed == () + + +@pytest.mark.asyncio +@pytest.mark.parametrize("deferred_for_a_post_call_guardrail", [False, True]) +async def test_stream_auto_execute_hung_up_during_the_follow_up_bills_round_one_and_the_partial_answer_on_one_row( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch, deferred_for_a_post_call_guardrail: bool +): + from litellm.proxy.common_request_processing import ProxyBaseLLMRequestProcessing + + monkeypatch.setattr(litellm, "cost_margin_config", {"openai": {"fixed_amount": _FIXED_MARGIN}}) + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _streamed(_CHAT_ROUND_TWO))) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-hung-up-mid-answer", stream=True) + + with _echo_tool_served(): + stream: Final = await _open_chat_echo_stream(logging_obj) + if deferred_for_a_post_call_guardrail: + ProxyBaseLLMRequestProcessing(data={})._arm_deferred_stream_dispatch( + response=stream, route_type="acompletion", user_api_key_dict=MagicMock(), logging_obj=logging_obj + ) + await _read_until(stream, "PING") + await ProxyBaseLLMRequestProcessing.finalize_streaming_generator_cleanup( + request=None, request_data={"litellm_logging_obj": logging_obj}, response=stream, client_disconnected=True + ) + await recorder.settled(logged=1) + + _assert_one_row_bills_round_one_and_the_partial_answer(recorder, respx_mock) + + +def _chat_rounds_router() -> litellm.Router: + return litellm.Router( + model_list=[ + { + "model_name": "mcp-chat-rounds", + "litellm_params": { + "model": "openai/gpt-5.6", + "api_key": "sk-mcp-chat-rounds", + "api_base": "https://mcp-chat-rounds.example.com/v1", + }, + } + ] + ) + + +class _StallsOnceTheAnswerStarts(httpx.AsyncByteStream): + async def __aiter__(self) -> AsyncIterator[bytes]: + yield _sse(_chat_stream_events(_CHAT_ROUND_TWO)[:1]) + await asyncio.Event().wait() + + +def _streams_content(body: object, content: str) -> bool: + lines: Final = body.decode().splitlines() if isinstance(body, bytes) else () + events: Final = tuple(json.loads(line.removeprefix("data: ")) for line in lines if line.startswith("data: {")) + return any(event["choices"] and event["choices"][0]["delta"].get("content") == content for event in events) + + +async def _serve_until_the_client_hangs_up_after(response: StreamingResponse, content: str) -> None: + read: Final = asyncio.Event() + + async def receive() -> Message: + await read.wait() + return {"type": "http.disconnect"} + + async def send(message: Message) -> None: + if message["type"] == "http.response.body" and _streams_content(message.get("body"), content): + read.set() + + uvicorn_scope: Final = {"type": "http", "asgi": {"version": "3.0", "spec_version": "2.3"}, "headers": []} + await response(uvicorn_scope, receive, send) + + +@pytest.mark.asyncio +@pytest.mark.parametrize("deferred_for_a_post_call_guardrail", [False, True]) +async def test_stream_auto_execute_routed_through_the_proxy_bills_the_partial_answer_when_the_client_hangs_up( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch, deferred_for_a_post_call_guardrail: bool +): + from litellm.proxy import proxy_server + from litellm.proxy._types import UserAPIKeyAuth + from litellm.proxy.common_request_processing import ProxyBaseLLMRequestProcessing, create_response + + monkeypatch.setattr(litellm, "cost_margin_config", {"openai": {"fixed_amount": _FIXED_MARGIN}}) + _serve_chat_rounds( + respx_mock, + monkeypatch, + ( + _streamed(_CHAT_ROUND_ONE), + Response(200, stream=_StallsOnceTheAnswerStarts(), headers={"content-type": "text/event-stream"}), + ), + ) + router: Final = _chat_rounds_router() + monkeypatch.setattr(proxy_server, "llm_router", router) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-routed-hang-up", stream=True) + + with _echo_tool_served(): + stream: Final = await router.acompletion( + model="mcp-chat-rounds", + messages=[{"role": "user", "content": "Call echo with PING"}], + tools=[{"type": "mcp", "server_url": "litellm_proxy/mcp/demo", "require_approval": "never"}], + stream=True, + stream_options={"include_usage": True}, + litellm_logging_obj=logging_obj, + ) + if deferred_for_a_post_call_guardrail: + ProxyBaseLLMRequestProcessing(data={})._arm_deferred_stream_dispatch( + response=stream, route_type="acompletion", user_api_key_dict=MagicMock(), logging_obj=logging_obj + ) + served: Final = await create_response( + proxy_server.async_data_generator( + response=stream, + user_api_key_dict=UserAPIKeyAuth(), + request_data={"litellm_logging_obj": logging_obj, "model": "mcp-chat-rounds", "metadata": {}}, + ), + "text/event-stream", + {}, + ) + assert isinstance(served, StreamingResponse) + await _serve_until_the_client_hangs_up_after(served, "PING") + await recorder.settled(logged=1) + + _assert_one_row_bills_round_one_and_the_partial_answer(recorder, respx_mock) + + +async def _stop_reading_once_the_proxy_sent(served: AsyncIterator[object], content: str) -> None: + async for frame in served: + if _streams_content(frame, content): + return + pytest.fail(f"the proxy never sent {content!r}") + + +@pytest.mark.asyncio +@pytest.mark.parametrize("deferred_for_a_post_call_guardrail", [False, True]) +async def test_stream_auto_execute_routed_through_the_proxy_bills_the_partial_answer_when_hung_up_between_reads( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch, deferred_for_a_post_call_guardrail: bool +): + from litellm.proxy import proxy_server + from litellm.proxy._types import UserAPIKeyAuth + from litellm.proxy.common_request_processing import ProxyBaseLLMRequestProcessing + + monkeypatch.setattr(litellm, "cost_margin_config", {"openai": {"fixed_amount": _FIXED_MARGIN}}) + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _streamed(_CHAT_ROUND_TWO))) + router: Final = _chat_rounds_router() + monkeypatch.setattr(proxy_server, "llm_router", router) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-routed-hang-up-between-reads", stream=True) + + with _echo_tool_served(): + stream: Final = await router.acompletion( + model="mcp-chat-rounds", + messages=[{"role": "user", "content": "Call echo with PING"}], + tools=[{"type": "mcp", "server_url": "litellm_proxy/mcp/demo", "require_approval": "never"}], + stream=True, + stream_options={"include_usage": True}, + litellm_logging_obj=logging_obj, + ) + if deferred_for_a_post_call_guardrail: + ProxyBaseLLMRequestProcessing(data={})._arm_deferred_stream_dispatch( + response=stream, route_type="acompletion", user_api_key_dict=MagicMock(), logging_obj=logging_obj + ) + served: Final = proxy_server.async_data_generator( + response=stream, + user_api_key_dict=UserAPIKeyAuth(), + request_data={"litellm_logging_obj": logging_obj, "model": "mcp-chat-rounds", "metadata": {}}, + ) + await _stop_reading_once_the_proxy_sent(served, "PING") + await served.aclose() + await recorder.settled(logged=1) + + _assert_one_row_bills_round_one_and_the_partial_answer(recorder, respx_mock) + + +_LONGEST_WAIT_FOR_THE_LOOP_TO_MOVE_SECONDS: Final = 10 + + +class _CountsTheToolResultWhileTheLoopRuns: + def __init__(self, loop: asyncio.AbstractEventLoop, count: Callable[..., int]) -> None: + self._loop: Final = loop + self._count: Final = count + self._counting: Final = asyncio.Event() + self._loop_moved: Final = threading.Event() + self.loop_moved_while_counting: tuple[bool, ...] = () + + def __call__(self, model: str | None = None, messages: list[object] | None = None, **kwargs: object) -> int: + if any(isinstance(message, Mapping) and message.get("role") == "tool" for message in messages or ()): + self._loop.call_soon_threadsafe(self._counting.set) + moved: Final = self._loop_moved.wait(timeout=_LONGEST_WAIT_FOR_THE_LOOP_TO_MOVE_SECONDS) + self.loop_moved_while_counting = (*self.loop_moved_while_counting, moved) + return self._count(model=model, messages=messages, **kwargs) + + async def keep_the_loop_moving(self) -> None: + await self._counting.wait() + self._loop_moved.set() + + +@contextmanager +def _partial_count_watched(monkeypatch: pytest.MonkeyPatch) -> Iterator[_CountsTheToolResultWhileTheLoopRuns]: + from litellm.litellm_core_utils import streaming_chunk_builder_utils + + counter: Final = _CountsTheToolResultWhileTheLoopRuns( + asyncio.get_running_loop(), streaming_chunk_builder_utils.token_counter + ) + monkeypatch.setattr(streaming_chunk_builder_utils, "token_counter", counter) + other_task: Final = asyncio.create_task(counter.keep_the_loop_moving()) + try: + yield counter + finally: + other_task.cancel() + + +class _StallsAfterTheAnswerStarts(httpx.AsyncByteStream): + def __init__(self) -> None: + self.stalled: Final = asyncio.Event() + + async def __aiter__(self) -> AsyncIterator[bytes]: + yield _sse(_chat_stream_events(_CHAT_ROUND_TWO)[:1]) + self.stalled.set() + await asyncio.Event().wait() + + +@pytest.mark.asyncio +async def test_stream_auto_execute_cancelled_mid_answer_counts_the_partial_off_the_loop_and_bills_it_once( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch +): + monkeypatch.setattr(litellm, "cost_margin_config", {"openai": {"fixed_amount": _FIXED_MARGIN}}) + answer: Final = _StallsAfterTheAnswerStarts() + _serve_chat_rounds( + respx_mock, + monkeypatch, + (_streamed(_CHAT_ROUND_ONE), Response(200, stream=answer, headers={"content-type": "text/event-stream"})), + ) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-cancelled-off-the-loop", stream=True) + + with _echo_tool_served(), _partial_count_watched(monkeypatch) as counter: + stream: Final = await _open_chat_echo_stream(logging_obj) + reading: Final = asyncio.create_task(_read_all(stream)) + await asyncio.wait_for(answer.stalled.wait(), timeout=10) + reading.cancel() + with pytest.raises(asyncio.CancelledError): + await reading + await stream.aclose() + await recorder.settled(logged=1) + + assert counter.loop_moved_while_counting == (True,) + _assert_one_row_bills_round_one_and_the_partial_answer(recorder, respx_mock) + + +@pytest.mark.asyncio +async def test_stream_auto_execute_routed_through_the_proxy_and_closed_between_reads_counts_the_partial_off_the_loop( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch +): + from litellm.proxy import proxy_server + from litellm.proxy._types import UserAPIKeyAuth + + monkeypatch.setattr(litellm, "cost_margin_config", {"openai": {"fixed_amount": _FIXED_MARGIN}}) + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _streamed(_CHAT_ROUND_TWO))) + router: Final = _chat_rounds_router() + monkeypatch.setattr(proxy_server, "llm_router", router) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-routed-closed-off-the-loop", stream=True) + + with _echo_tool_served(), _partial_count_watched(monkeypatch) as counter: + stream: Final = await router.acompletion( + model="mcp-chat-rounds", + messages=[{"role": "user", "content": "Call echo with PING"}], + tools=[{"type": "mcp", "server_url": "litellm_proxy/mcp/demo", "require_approval": "never"}], + stream=True, + stream_options={"include_usage": True}, + litellm_logging_obj=logging_obj, + ) + served: Final = proxy_server.async_data_generator( + response=stream, + user_api_key_dict=UserAPIKeyAuth(), + request_data={"litellm_logging_obj": logging_obj, "model": "mcp-chat-rounds", "metadata": {}}, + ) + await _stop_reading_once_the_proxy_sent(served, "PING") + await served.aclose() + await recorder.settled(logged=1) + + assert counter.loop_moved_while_counting == (True,) + _assert_one_row_bills_round_one_and_the_partial_answer(recorder, respx_mock) + + +class _StallsBeforeTheAnswerStarts(httpx.AsyncByteStream): + def __init__(self) -> None: + self.opened: Final = asyncio.Event() + + async def __aiter__(self) -> AsyncIterator[bytes]: + yield b": the answer has not started\n\n" + self.opened.set() + await asyncio.Event().wait() + + +@pytest.mark.asyncio +@pytest.mark.parametrize("deferred_for_a_post_call_guardrail", [False, True]) +async def test_stream_auto_execute_hung_up_before_the_answer_starts_bills_round_one_alone_on_one_row( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch, deferred_for_a_post_call_guardrail: bool +): + from litellm.proxy.common_request_processing import ProxyBaseLLMRequestProcessing + + answer: Final = _StallsBeforeTheAnswerStarts() + _serve_chat_rounds( + respx_mock, + monkeypatch, + (_streamed(_CHAT_ROUND_ONE), Response(200, stream=answer, headers={"content-type": "text/event-stream"})), + ) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-hung-up-before-answer", stream=True) + + with _echo_tool_served(): + stream: Final = await _open_chat_echo_stream(logging_obj) + if deferred_for_a_post_call_guardrail: + ProxyBaseLLMRequestProcessing(data={})._arm_deferred_stream_dispatch( + response=stream, route_type="acompletion", user_api_key_dict=MagicMock(), logging_obj=logging_obj + ) + reading: Final = asyncio.create_task(_read_all(stream)) + await asyncio.wait_for(answer.opened.wait(), timeout=10) + reading.cancel() + with pytest.raises(asyncio.CancelledError): + await reading + await ProxyBaseLLMRequestProcessing.finalize_streaming_generator_cleanup( + request=None, request_data={"litellm_logging_obj": logging_obj}, response=stream, client_disconnected=True + ) + await recorder.settled(logged=1) + + _assert_one_row_bills_round_one(recorder.logged) + assert recorder.failed == () + + +async def _read_through_the_usage_chunk( + chunks: AsyncIterator[ModelResponseStream], read: tuple[ModelResponseStream, ...] = () +) -> tuple[ModelResponseStream, ...]: + chunk: Final = await anext(chunks) + if isinstance(getattr(chunk, "usage", None), Usage): + return (*read, chunk) + return await _read_through_the_usage_chunk(chunks, (*read, chunk)) + + +@pytest.mark.asyncio +@pytest.mark.parametrize("deferred_for_a_post_call_guardrail", [False, True]) +async def test_stream_auto_execute_hung_up_after_its_whole_answer_bills_both_rounds_once( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch, deferred_for_a_post_call_guardrail: bool +): + from litellm.proxy.common_request_processing import ProxyBaseLLMRequestProcessing + + monkeypatch.setattr(litellm, "cost_margin_config", {"openai": {"fixed_amount": _FIXED_MARGIN}}) + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _streamed(_CHAT_ROUND_TWO))) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-hung-up-after-answer", stream=True) + + with _echo_tool_served(): + stream: Final = await _open_chat_echo_stream(logging_obj) + if deferred_for_a_post_call_guardrail: + ProxyBaseLLMRequestProcessing(data={})._arm_deferred_stream_dispatch( + response=stream, route_type="acompletion", user_api_key_dict=MagicMock(), logging_obj=logging_obj + ) + chunks: Final = await _read_through_the_usage_chunk(aiter(stream)) + await ProxyBaseLLMRequestProcessing.finalize_streaming_generator_cleanup( + request=None, request_data={"litellm_logging_obj": logging_obj}, response=stream, client_disconnected=True + ) + await recorder.settled(logged=1) + + _assert_one_stream_row_bills_both_rounds_once(recorder.logged, chunks) + assert recorder.failed == () + + +@pytest.mark.asyncio +async def test_stream_auto_execute_bridged_to_messages_bills_round_one_when_its_follow_up_breaks( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch +): + from litellm.proxy.common_request_processing import ProxyBaseLLMRequestProcessing + + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _cut_once_the_answer_starts())) + recorder: Final = _ChatRoundRecorder() + logging_obj: Final = _chat_rounds_logging_obj(recorder, "mcp-chat-stream-bridged-to-messages", stream=True) + + async def bridged(stream: CustomStreamWrapper) -> AsyncIterator[ModelResponseStream]: + async for chunk in stream: + yield chunk + + with _echo_tool_served(): + relayed: Final = bridged(await _open_chat_echo_stream(logging_obj)) + ProxyBaseLLMRequestProcessing(data={})._arm_deferred_stream_dispatch( + response=relayed, route_type="anthropic_messages", user_api_key_dict=MagicMock(), logging_obj=logging_obj + ) + with pytest.raises(litellm.ServiceUnavailableError): + _ = [chunk async for chunk in relayed] + await recorder.settled(logged=1, failed=1) + + _assert_one_row_bills_round_one(recorder.logged) + + +def _read_synchronously_until_the_follow_up_answers( + logging_obj: Logging, loop: asyncio.AbstractEventLoop, before_dropping: Callable[[], None] = lambda: None +) -> None: + stream: Final = loop.run_until_complete(_open_chat_echo_stream(logging_obj)) + for chunk in stream: + if chunk.choices and chunk.choices[0].delta.content == "PING": + before_dropping() + return + pytest.fail("the stream never sent 'PING'") + + +_CHAT_STREAM_BILLING_NAMES: Final = ("MCPStreamingIterator", "ChatStreamRoundBilling", "_log_success") + + +def _names_chat_stream_billing(text: str) -> bool: + return any(name in text for name in _CHAT_STREAM_BILLING_NAMES) + + +class _UnraisableRecorder: + def __init__(self) -> None: + self.raised: tuple[str, ...] = () + + def __call__(self, unraisable: "sys.UnraisableHookArgs") -> None: + self.raised = (*self.raised, f"{unraisable.object!r}: {unraisable.exc_value!r}") + + +def test_stream_auto_execute_read_synchronously_and_dropped_mid_answer_bills_the_finished_first_round( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch +): + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _streamed(_CHAT_ROUND_TWO))) + recorder: Final = _ChatRoundRecorder() + unraisable: Final = _UnraisableRecorder() + loop: Final = asyncio.new_event_loop() + gc.collect() + monkeypatch.setattr(sys, "unraisablehook", unraisable) + asyncio.set_event_loop(loop) + + try: + with _echo_tool_served(), warnings.catch_warnings(record=True) as caught: + warnings.simplefilter("always") + _read_synchronously_until_the_follow_up_answers( + _chat_rounds_logging_obj(recorder, "mcp-chat-stream-sync-dropped", stream=True), loop + ) + gc.collect() + loop.run_until_complete(recorder.settled(logged=1)) + finally: + asyncio.set_event_loop(None) + loop.close() + + _assert_one_row_bills_round_one(recorder.logged) + assert [raised for raised in unraisable.raised if _names_chat_stream_billing(raised)] == [] + assert [str(warning.message) for warning in caught if _names_chat_stream_billing(str(warning.message))] == [] + + +def test_stream_auto_execute_read_synchronously_and_dropped_after_its_loop_closed_drops_without_errors( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch +): + _serve_chat_rounds(respx_mock, monkeypatch, (_streamed(_CHAT_ROUND_ONE), _streamed(_CHAT_ROUND_TWO))) + recorder: Final = _ChatRoundRecorder() + unraisable: Final = _UnraisableRecorder() + loop: Final = asyncio.new_event_loop() + gc.collect() + monkeypatch.setattr(sys, "unraisablehook", unraisable) + asyncio.set_event_loop(loop) + + try: + with _echo_tool_served(), warnings.catch_warnings(record=True) as caught: + warnings.simplefilter("always") + _read_synchronously_until_the_follow_up_answers( + _chat_rounds_logging_obj(recorder, "mcp-chat-stream-sync-dropped-after-close", stream=True), + loop, + before_dropping=loop.close, + ) + gc.collect() + finally: + asyncio.set_event_loop(None) + loop.close() + + assert [raised for raised in unraisable.raised if _names_chat_stream_billing(raised)] == [] + assert [str(warning.message) for warning in caught if _names_chat_stream_billing(str(warning.message))] == [] diff --git a/tests/unit/responses/mcp/test_parked_stream_dispatch.py b/tests/unit/responses/mcp/test_parked_stream_dispatch.py new file mode 100644 index 00000000000..dd380c582c8 --- /dev/null +++ b/tests/unit/responses/mcp/test_parked_stream_dispatch.py @@ -0,0 +1,17 @@ +from typing import Final + +import pytest + +from litellm.responses.mcp.parked_stream_dispatch import parked_success +from litellm.types.utils import ModelResponse + +_ASSEMBLED: Final = ModelResponse(id="chatcmpl-parked", model="gpt-5.6") + + +@pytest.mark.parametrize( + "parked", + [(), (_ASSEMBLED,), (_ASSEMBLED, False, "not a logger"), (_ASSEMBLED, False, None, None)], + ids=("empty", "assembled-alone", "third-part-is-not-a-logger", "four-parts"), +) +def test_a_parked_dispatch_of_an_unknown_shape_is_never_dispatched(parked: tuple[object, ...]): + assert parked_success(parked) is None From 4786dd861b1b8c001117e0403cc2c606c193896c Mon Sep 17 00:00:00 2001 From: zoveki <349028329+zoveki@users.noreply.github.com> Date: Fri, 30 Oct 2026 07:36:01 -0700 Subject: [PATCH 04/18] fix(proxy)!: scope /health by wildcard keys and team model allowlists the way auth does (#49155) * test(proxy): cover /health scoping for wildcard keys and team model allowlists * fix(proxy): scope /health with the model access check auth runs on a request * fix(proxy): scope /health to the model group an allowed name routes to A key alias, team alias, or router model_group_alias lets auth admit a requested name by checking its target, so /health also kept the deployment under the requested name, one the key's requests never reach. Keep only the served model group each admitted name dispatches to, through the same alias layers /v1/models mirrors, and let alias_map read its own read-only output. * test(proxy): cover /health skipping a deployment that has no model name for a restricted key * fix(health): show deployments reached through wildcard and alias allowlists, and by id when an alias shadows the name * chore(router): drop should_include_deployment from the outside-use allowlist now that /health no longer calls it * fix(health): never list a default-fallback deployment for an unserved grant, and keep access-group keys to their group's deployments * fix(health): resolve ?model= through the caller's aliases the way chat does * fix(proxy): apply user and agent model limits to /health and yield between checks /health now runs the same user and agent model checks chat runs, so a personal key whose owner may call only some models, or an agent key limited by its access group, no longer sees or probes other deployments. Each per-name check yields to the event loop, and ?model= reaches the wildcard deployment serving an allowed name * fix(proxy): check owner and agent grants for wildcard routes in /health and fail on non-access auth errors * fix(proxy): answer /health auth lookup failures the way chat does and record ?model= checks under the requested name * test(proxy): pass recorded_model_name in the direct health save tests * fix(proxy): count alias names as /health candidates so aliased names a key may call get probed * fix(health): narrow wildcard-served deployments to the key's access groups * fix(health): narrow every deployment serving a name to the key's access groups, team-owned ones included * fix(health): narrow a team deployment's internal name to the key's access groups like its public name * fix(health): keep a team deployment's internal name narrow when a limit holds only that name An agent or key limit that holds one team deployment's internal name no longer reaches every deployment behind the team's public name. The internal name maps to its public name only when the public name is also admitted. * test(health): cover all-proxy-models next to * for a key that also holds an access group * refactor(auth): keep the _as_proxy_exception name so main's tests still find it * fix(health): keep an internal team name a wildcard grant admits as its own target --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> --- litellm/proxy/auth/auth_checks.py | 47 +- .../proxy/common_utils/model_listing_utils.py | 11 +- .../health_endpoints/_health_endpoints.py | 427 ++++++++++- ...st_auth_checks_object_access_and_lookup.py | 10 + .../health_endpoints/test_health_endpoints.py | 720 +++++++++++++++++- .../unit/proxy/test_health_check_functions.py | 2 + 6 files changed, 1164 insertions(+), 53 deletions(-) diff --git a/litellm/proxy/auth/auth_checks.py b/litellm/proxy/auth/auth_checks.py index b8d9530d2eb..f80296f4bd7 100644 --- a/litellm/proxy/auth/auth_checks.py +++ b/litellm/proxy/auth/auth_checks.py @@ -84,6 +84,7 @@ from litellm.proxy._types import ( UserAPIKeyAuth, ) from litellm.proxy.agent_endpoints.auth.agent_access_groups import ( + AgentAccessGroupCeiling, CeilingResolver, resolve_agent_access_group_ceiling, ) @@ -1684,6 +1685,7 @@ def get_actual_routes(allowed_routes: list) -> list: KEY_END_USER_BUDGET_ID_METADATA_FIELD: Final = "end_user_budget_id" _KEY_METADATA_ADAPTER: Final[TypeAdapter[Mapping[str, object] | None]] = TypeAdapter(Mapping[str, object] | None) +_OBJECTS_ADAPTER: Final[TypeAdapter[tuple[object, ...]]] = TypeAdapter(tuple[object, ...]) def get_key_end_user_budget_id(key_metadata: Mapping[str, object] | None) -> str | None: @@ -5085,6 +5087,36 @@ async def can_agent_call_model( ) +async def _agent_access_group_ceilings( + valid_token: UserAPIKeyAuth, resolve_ceiling: CeilingResolver +) -> tuple[AgentAccessGroupCeiling, ...]: + if not valid_token.agent_id: + return () + + from litellm.proxy.agent_endpoints.auth.agent_access_groups import resolve_managed_agent_ceilings + + managed: Final = managed_agent_policy(valid_token) + if managed is not None: + return await resolve_managed_agent_ceilings(managed) + unmanaged: Final = await resolve_ceiling(valid_token.agent_id) + return (unmanaged,) if unmanaged is not None else () + + +def _managed_agent_model_grants(valid_token: UserAPIKeyAuth) -> frozenset[str]: + managed: Final = managed_agent_policy(valid_token) + granted: Final = (managed.object_permission or {}).get("models") if managed is not None else None + if not isinstance(granted, (list, tuple)): + return frozenset() + return frozenset(name for name in _OBJECTS_ADAPTER.validate_python(granted) if isinstance(name, str)) + + +async def agent_model_grants( + valid_token: UserAPIKeyAuth, resolve_ceiling: CeilingResolver = resolve_agent_access_group_ceiling +) -> frozenset[str]: + ceilings: Final = await _agent_access_group_ceilings(valid_token, resolve_ceiling) + return _managed_agent_model_grants(valid_token).union(*(ceiling.models for ceiling in ceilings)) + + async def _check_agent_access_group_model_access( model: str | list[str] | None, # mutable-ok: _can_object_call_model and the client message helper take list[str] valid_token: UserAPIKeyAuth | None, @@ -5092,20 +5124,9 @@ async def _check_agent_access_group_model_access( resolve_ceiling: CeilingResolver = resolve_agent_access_group_ceiling, ) -> Literal[True]: """Attached groups naming no model deny every model; the empty allowlist in ``_can_object_call_model`` allows.""" - if not model or valid_token is None or not valid_token.agent_id: + if not model or valid_token is None: return True - - from litellm.proxy.agent_endpoints.auth.agent_access_groups import resolve_managed_agent_ceilings - - managed: Final = managed_agent_policy(valid_token) - unmanaged: Final = await resolve_ceiling(valid_token.agent_id) if managed is None else None - ceilings: Final = ( - await resolve_managed_agent_ceilings(managed) - if managed is not None - else (unmanaged,) - if unmanaged is not None - else () - ) + ceilings: Final = await _agent_access_group_ceilings(valid_token, resolve_ceiling) dispatched: Final = _resolve_team_alias( model, team_model_aliases_for_auth_check(valid_token), valid_token.team_id, llm_router ) diff --git a/litellm/proxy/common_utils/model_listing_utils.py b/litellm/proxy/common_utils/model_listing_utils.py index ad4c3bd8770..a9424018d9a 100644 --- a/litellm/proxy/common_utils/model_listing_utils.py +++ b/litellm/proxy/common_utils/model_listing_utils.py @@ -219,7 +219,7 @@ def caller_alias_maps( def alias_map(aliases: object) -> Mapping[str, str]: try: - entries: Final = _ALIAS_ENTRIES.validate_python(aliases, strict=True) + entries: Final = _ALIAS_ENTRIES.validate_python(aliases) except ValidationError: return _NO_ALIASES return MappingProxyType( @@ -287,6 +287,15 @@ def _dispatch(model_id: str, layers: Sequence[_AliasLayer]) -> tuple[str, bool]: return reduce(_through_layer, layers, (model_id, False)) +def dispatched_model(model_id: str, aliases: CallerAliases) -> str: + target, _ = _dispatch(model_id, _alias_layers(aliases)) + return target + + +def requestable_alias_names(aliases: CallerAliases) -> frozenset[str]: + return frozenset(chain.from_iterable(layer.aliases for layer in _alias_layers(aliases))) + + def _lookup_id(name: str, lookup_by_response: Mapping[str, str], lookup_ids: Container[str]) -> str | None: if name in lookup_by_response: return lookup_by_response[name] diff --git a/litellm/proxy/health_endpoints/_health_endpoints.py b/litellm/proxy/health_endpoints/_health_endpoints.py index 59c63a18f0f..8545e0609db 100644 --- a/litellm/proxy/health_endpoints/_health_endpoints.py +++ b/litellm/proxy/health_endpoints/_health_endpoints.py @@ -7,13 +7,15 @@ import secrets import time import traceback from collections.abc import Iterable, Mapping, Sequence +from dataclasses import dataclass from datetime import datetime, timedelta, timezone +from itertools import chain, groupby from types import MappingProxyType from typing import Final, Literal, TypedDict import fastapi from fastapi import APIRouter, Depends, HTTPException, Request, Response, status -from pydantic import TypeAdapter +from pydantic import BaseModel, ConfigDict, TypeAdapter from typing_extensions import ReadOnly import litellm @@ -31,7 +33,9 @@ from litellm.proxy._types import ( CallInfo, EnterpriseLicenseData, Litellm_EntityType, + LiteLLM_UserTable, LitellmUserRoles, + ModelAccessDeniedProxyException, ProxyException, SpecialModelNames, UserAPIKeyAuth, @@ -39,13 +43,28 @@ from litellm.proxy._types import ( ) from litellm.proxy.auth.auth_checks import ( _resolve_key_models_for_auth_check, # pyright: ignore[reportPrivateUsage] # the auth layer's sentinel resolution, reused so /health scopes exactly like a request + agent_model_grants, + can_agent_call_model, + can_key_call_resolved_model, + can_user_call_model, + get_user_object, + key_model_aliases_for_auth_check, + model_matches_any_wildcard_pattern_in_list, + team_model_aliases_for_auth_check, +) +from litellm.proxy.auth.auth_exception_handler import ( + _as_proxy_exception, # pyright: ignore[reportPrivateUsage] # one auth-failure mapping, shared with user_api_key_auth ) from litellm.proxy.auth.auth_utils import ( _BANNED_REQUEST_BODY_PARAMS, # pyright: ignore[reportPrivateUsage] # one canonical list, shared with the request-body check ) -from litellm.proxy.auth.model_checks import get_key_models from litellm.proxy.auth.user_api_key_auth import user_api_key_auth from litellm.proxy.common_utils import http_parsing_utils +from litellm.proxy.common_utils.model_listing_utils import ( + CallerAliases, + dispatched_model, + requestable_alias_names, +) from litellm.proxy.common_utils.openai_error_payload import proxy_exception_for from litellm.proxy.db.db_lookup_gate import db_lookup_stall_tracker from litellm.proxy.db.exception_handler import PrismaDBExceptionHandler @@ -79,6 +98,7 @@ from litellm.router_utils.clientside_credential_handler import ( _ADMIN_CONFIG_FIELDS_TO_CLEAR_ON_BASE_OVERRIDE, # pyright: ignore[reportPrivateUsage] # one canonical list, shared with the router path clientside_credential_keys, ) +from litellm.router_utils.common_utils import resolve_model_group_alias from litellm.secret_managers.main import get_secret_bool from litellm.types.proxy.litellm_pre_call_utils import RedactedDict, SecretFields from litellm.types.router import reject_server_owned_wif_fields @@ -253,6 +273,28 @@ def get_callback_identifier(callback): router: Final = APIRouter() _OBJECT_MAPPING: Final = TypeAdapter(Mapping[str, object]) +_OBJECTS: Final = TypeAdapter(tuple[object, ...]) +_DEPLOYMENTS: Final = TypeAdapter(tuple[Mapping[str, object], ...]) + + +class _ModelGrants(BaseModel): + model_config = ConfigDict(frozen=True, from_attributes=True) + + models: tuple[object, ...] = () + + +class _ModelInfoAccessGroups(BaseModel): + model_config = ConfigDict(frozen=True) + + access_groups: tuple[object, ...] + + +class _DeploymentAccessGroups(BaseModel): + model_config = ConfigDict(frozen=True) + + model_info: _ModelInfoAccessGroups + + services = ( Literal[ "slack_budget_alerts", @@ -1045,42 +1087,339 @@ def _strip_admin_only_fields_from_health_result(result: dict) -> dict: return out -def _health_accessible_model_names( - user_api_key_dict: UserAPIKeyAuth, llm_router: Router | None -) -> frozenset[str] | None: - """Model names the caller may health-check, or None when the key is unrestricted.""" +def _key_models_unrestricted(user_api_key_dict: UserAPIKeyAuth) -> bool: granted_models: Final = _resolve_key_models_for_auth_check(user_api_key_dict) - if not granted_models or SpecialModelNames.all_proxy_models.value in granted_models: + return not granted_models or SpecialModelNames.all_proxy_models.value in granted_models + + +def _own_team_names(deployment: object, team_id: str | None) -> tuple[str, str] | None: + match deployment: + case { + "model_name": str(internal_name), + "model_info": {"team_id": str(owner_team_id), "team_public_model_name": str(public_name)}, + } if owner_team_id == team_id: + return (internal_name, public_name) + case _: + return None + + +def _names_caller_may_request(deployment: object, team_id: str | None) -> frozenset[str]: + team_names: Final = _own_team_names(deployment, team_id) + if team_names is not None: + return frozenset(team_names) + match deployment: + case {"model_name": str(name)}: + return frozenset((name,)) + case _: + return frozenset() + + +def _public_names_by_internal_name(llm_model_list: Sequence[object], team_id: str | None) -> Mapping[str, str]: + team_names: Final = (_own_team_names(deployment, team_id) for deployment in llm_model_list) + return MappingProxyType(dict(names for names in team_names if names is not None)) + + +async def _key_owner(user_api_key_dict: UserAPIKeyAuth) -> LiteLLM_UserTable | None: + if user_api_key_dict.team_id is not None or user_api_key_dict.user_id is None: return None + from litellm.proxy.proxy_server import prisma_client, proxy_logging_obj, user_api_key_cache + + try: + return await get_user_object( + user_id=user_api_key_dict.user_id, + prisma_client=prisma_client, + user_api_key_cache=user_api_key_cache, + user_id_upsert=False, + parent_otel_span=user_api_key_dict.parent_otel_span, + proxy_logging_obj=proxy_logging_obj, + ) + except Exception: # noqa: BLE001 # auth reads an unloadable user as no user limit, so /health does too + return None + + +async def _key_may_call_model( + model_name: str, + user_api_key_dict: UserAPIKeyAuth, + owner: LiteLLM_UserTable | None, + llm_router: Router | None, +) -> bool: + from litellm.proxy.proxy_server import prisma_client, proxy_logging_obj, user_api_key_cache + + await asyncio.sleep(0) + try: + await can_key_call_resolved_model( + model=model_name, llm_model_list=None, valid_token=user_api_key_dict, llm_router=llm_router + ) + await can_agent_call_model( + model=model_name, + valid_token=user_api_key_dict, + llm_router=llm_router, + prisma_client=prisma_client, + user_api_key_cache=user_api_key_cache, + proxy_logging_obj=proxy_logging_obj, + ) + await can_user_call_model( + model=model_name, + llm_router=llm_router, + user_object=owner, + key_model_aliases=key_model_aliases_for_auth_check(user_api_key_dict), + ) + except ModelAccessDeniedProxyException: + return False + except ProxyException: + raise + except Exception as failure: + if isinstance(failure, HTTPException) and failure.status_code == status.HTTP_403_FORBIDDEN: + return False + raise _as_proxy_exception(failure) from failure + return True + + +async def _caller_aliases(user_api_key_dict: UserAPIKeyAuth) -> CallerAliases: + from litellm.proxy.proxy_server import prisma_client, proxy_config, proxy_logging_obj + + router_settings: Final = await proxy_config.get_hierarchical_router_settings( + user_api_key_dict, prisma_client, proxy_logging_obj + ) + return CallerAliases( + team_model_aliases_for_auth_check(user_api_key_dict), + key_model_aliases_for_auth_check(user_api_key_dict), + router_settings.get("model_group_alias") if router_settings is not None else None, + ) + + +def _model_auth_checks(model_name: str, aliases: CallerAliases) -> str: + return resolve_model_group_alias(aliases.router, model_name) or model_name + + +def _deployment_id(deployment: object) -> str | None: + match deployment: + case {"model_info": {"id": str(model_id)}}: + return model_id + case _: + return None + + +def _granted_names(user_api_key_dict: UserAPIKeyAuth) -> frozenset[str]: + grants: Final = ( + *_ModelGrants.model_validate(user_api_key_dict).models, + *_OBJECTS.validate_python(user_api_key_dict.team_models), + ) + return frozenset(name for name in grants if isinstance(name, str)) + + +async def _names_any_limit_grants(user_api_key_dict: UserAPIKeyAuth, owner: LiteLLM_UserTable | None) -> frozenset[str]: + owner_grants: Final = _ModelGrants.model_validate(owner).models if owner is not None else () + return ( + _granted_names(user_api_key_dict) + | frozenset(name for name in owner_grants if isinstance(name, str)) + | await agent_model_grants(user_api_key_dict) + ) + + +def _alias_names_caller_may_request(aliases: CallerAliases, llm_router: Router | None) -> frozenset[str]: + router_aliases: Final = frozenset(llm_router.registry().aliases()) if llm_router is not None else frozenset[str]() + return requestable_alias_names(aliases) | router_aliases + + +def _served_group(model_name: str, aliases: CallerAliases, llm_router: Router | None) -> str: + dispatched: Final = dispatched_model(model_name, aliases) + return llm_router.registry().resolve(dispatched) if llm_router is not None else dispatched + + +def _name_id_pairs(deployment: object, team_id: str | None) -> tuple[tuple[str, str], ...]: + model_id: Final = _deployment_id(deployment) + if model_id is None: + return () + return tuple((name, model_id) for name in _names_caller_may_request(deployment, team_id) | {model_id}) + + +def _ids_by_requested_name(llm_model_list: Sequence[object], team_id: str | None) -> Mapping[str, frozenset[str]]: + pairs: Final = sorted(chain.from_iterable(_name_id_pairs(deployment, team_id) for deployment in llm_model_list)) + return MappingProxyType( + {name: frozenset(model_id for _, model_id in group) for name, group in groupby(pairs, key=lambda pair: pair[0])} + ) + + +def _ids_of(deployments: Iterable[object]) -> frozenset[str]: + return frozenset(model_id for model_id in map(_deployment_id, deployments) if model_id is not None) + + +def _access_groups_of(deployment: object) -> frozenset[str]: + match deployment: + case {"model_info": {"access_groups": list()}}: + access_groups: Final = _DeploymentAccessGroups.model_validate(deployment).model_info.access_groups + return frozenset(access_group for access_group in access_groups if isinstance(access_group, str)) + case _: + return frozenset() + + +@dataclass(frozen=True, slots=True) +class _AccessGroupGrants: + granted_names: frozenset[str] + ids_in_granted_groups: frozenset[str] + + def requested_name(self, name: str, public_names: Mapping[str, str], admitted_names: frozenset[str]) -> str: + public_name: Final = public_names.get(name) + if public_name is None or public_name not in admitted_names or self._grants(name): + return name + return public_name + + def _grants(self, name: str) -> bool: + return name in self.granted_names or model_matches_any_wildcard_pattern_in_list(name, tuple(self.granted_names)) + + def narrow(self, serving_ids: frozenset[str], served_names: frozenset[str]) -> frozenset[str]: + in_granted_groups: Final = serving_ids & self.ids_in_granted_groups + granted_directly: Final = not self.granted_names.isdisjoint(served_names) + return serving_ids if granted_directly or not in_granted_groups else in_granted_groups + + +def _access_group_grants(llm_model_list: Sequence[object], user_api_key_dict: UserAPIKeyAuth) -> _AccessGroupGrants: + granted: Final = _granted_names(user_api_key_dict) + in_granted_groups: Final = ( + deployment for deployment in llm_model_list if not granted.isdisjoint(_access_groups_of(deployment)) + ) + return _AccessGroupGrants(granted, _ids_of(in_granted_groups)) + + +def _ids_serving_group( + group: str, + ids_by_name: Mapping[str, frozenset[str]], + user_api_key_dict: UserAPIKeyAuth, + llm_router: Router | None, + access_group_grants: _AccessGroupGrants, +) -> frozenset[str]: + named: Final = ids_by_name.get(group) + if named is not None: + return access_group_grants.narrow(named, frozenset((group,))) if llm_router is None: - return frozenset(granted_models) - registry: Final = llm_router.registry() + return frozenset() + team_id: Final = user_api_key_dict.team_id + wildcard_served: Final = llm_router.registry().deployments(group, source="wildcard", team_id=team_id) + served_names: Final = frozenset((group,)).union( + *(_names_caller_may_request(deployment, team_id) for deployment in wildcard_served) + ) + return access_group_grants.narrow(_ids_of(wildcard_served), served_names) + + +async def _names_key_may_request( + names: Iterable[str], + aliases: CallerAliases, + user_api_key_dict: UserAPIKeyAuth, + owner: LiteLLM_UserTable | None, + llm_router: Router | None, +) -> frozenset[str]: return frozenset( - get_key_models( - user_api_key_dict=user_api_key_dict, - proxy_model_list=registry.names(team_id=user_api_key_dict.team_id), - model_access_groups=registry.access_groups(), + [ + name + for name in names + if await _key_may_call_model(_model_auth_checks(name, aliases), user_api_key_dict, owner, llm_router) + ] + ) + + +def _ids_serving( + groups: Iterable[str], + ids_by_name: Mapping[str, frozenset[str]], + user_api_key_dict: UserAPIKeyAuth, + llm_router: Router | None, + access_group_grants: _AccessGroupGrants, +) -> frozenset[str]: + return frozenset[str]().union( + *( + _ids_serving_group(group, ids_by_name, user_api_key_dict, llm_router, access_group_grants) + for group in groups ) ) -def _caller_may_probe_deployment( - deployment: Mapping[str, object], - allowed_models: frozenset[str] | None, +def _shadowed_ids( + llm_model_list: Sequence[object], admitted_names: frozenset[str], reached_ids: frozenset[str], team_id: str | None +) -> frozenset[str]: + ids_of_admitted_names: Final = ( + _deployment_id(deployment) + for deployment in llm_model_list + if not admitted_names.isdisjoint(_names_caller_may_request(deployment, team_id)) + ) + return frozenset(model_id for model_id in ids_of_admitted_names if model_id is not None) - reached_ids + + +@dataclass(frozen=True, slots=True) +class _ReachableDeployments: + model_ids: frozenset[str] + groups: frozenset[str] + + def include(self, deployment: object, team_id: str | None) -> bool: + model_id: Final = _deployment_id(deployment) + if model_id is None: + return not self.groups.isdisjoint(_names_caller_may_request(deployment, team_id)) + return model_id in self.model_ids + + +_NOTHING_REACHABLE: Final = _ReachableDeployments(frozenset(), frozenset()) + + +async def _deployments_key_may_reach( + llm_model_list: Sequence[object], + aliases: CallerAliases, + user_api_key_dict: UserAPIKeyAuth, + owner: LiteLLM_UserTable | None, llm_router: Router | None, +) -> _ReachableDeployments: + candidate_names: Final = (await _names_any_limit_grants(user_api_key_dict, owner)).union( + _alias_names_caller_may_request(aliases, llm_router), + *(_names_caller_may_request(deployment, user_api_key_dict.team_id) for deployment in llm_model_list), + ) + admitted_names: Final = await _names_key_may_request(candidate_names, aliases, user_api_key_dict, owner, llm_router) + access_group_grants: Final = _access_group_grants(llm_model_list, user_api_key_dict) + public_names: Final = _public_names_by_internal_name(llm_model_list, user_api_key_dict.team_id) + groups: Final = frozenset( + _served_group(access_group_grants.requested_name(name, public_names, admitted_names), aliases, llm_router) + for name in admitted_names + ) + ids_by_name: Final = _ids_by_requested_name(llm_model_list, user_api_key_dict.team_id) + reached_ids: Final = _ids_serving(groups, ids_by_name, user_api_key_dict, llm_router, access_group_grants) + admitted_ids: Final = await _names_key_may_request( + _shadowed_ids(llm_model_list, admitted_names, reached_ids, user_api_key_dict.team_id), + aliases, + user_api_key_dict, + owner, + llm_router, + ) + id_groups: Final = frozenset(_served_group(model_id, aliases, llm_router) for model_id in admitted_ids) + return _ReachableDeployments( + reached_ids | _ids_serving(id_groups, ids_by_name, user_api_key_dict, llm_router, access_group_grants), groups + ) + + +def _model_param_target( + model: str, + aliases: CallerAliases, + llm_model_list: Sequence[Mapping[str, object]], team_id: str | None, - caller_is_admin: bool, + llm_router: Router | None, +) -> str: + group: Final = _served_group(model, aliases, llm_router) + if llm_router is None or deployments_targeted_by_name(llm_model_list, group, team_id): + return group + patterns: Final = frozenset( + deployment.get("model_name") + for deployment in llm_router.registry().deployments(group, source="wildcard", team_id=team_id) + ) + match tuple(patterns): + case (str(pattern),): + return pattern + case _: + return group + + +def _caller_may_probe_deployment( + deployment: Mapping[str, object], reachable: _ReachableDeployments, team_id: str | None, caller_is_admin: bool ) -> bool: """Same deployment visibility rule as routing: another team's deployment is never in scope, team-less callers included.""" if not caller_is_admin and not deployment_usable_by_team(deployment, team_id): return False - if allowed_models is None: - return True - if llm_router is None: - return deployment.get("model_name") in allowed_models - model: Final = dict(deployment) - registry: Final = llm_router.registry() - return any(registry.includes(model, model_name=name, team_id=team_id) for name in allowed_models) + return reachable.include(deployment, team_id) def _resolve_targeted_model_ids( @@ -1137,6 +1476,7 @@ def _filter_health_check_results_by_model_ids(results: dict, allowed_model_ids: async def _perform_health_check_and_save( model_list, target_model, + recorded_model_name: str | None, cli_model, details, prisma_client, @@ -1159,8 +1499,7 @@ async def _perform_health_check_and_save( # Optionally save health check result to database (non-blocking) if prisma_client is not None: - # For CLI model, use cli_model name; for router models, use target_model - model_name_for_db: Final = cli_model if cli_model is not None else target_model + model_name_for_db: Final = cli_model if cli_model is not None else recorded_model_name if model_name_for_db is not None: asyncio.create_task( _save_health_check_to_db( @@ -1290,6 +1629,7 @@ async def health_endpoint( cli_result: Final = await _perform_health_check_and_save( model_list=[], target_model=None, + recorded_model_name=None, cli_model=user_model, details=health_check_details, prisma_client=prisma_client, @@ -1304,15 +1644,39 @@ async def health_endpoint( status_code=status.HTTP_500_INTERNAL_SERVER_ERROR, detail={"error": "Model list not initialized"}, ) - allowed_models: Final = _health_accessible_model_names(user_api_key_dict, llm_router) - restrict_to_allowed_models: Final = not is_admin or allowed_models is not None + restrict_to_allowed_models: Final = not is_admin or not _key_models_unrestricted(user_api_key_dict) + aliases: Final = await _caller_aliases(user_api_key_dict) if restrict_to_allowed_models else None + reachable: Final = ( + await _deployments_key_may_reach( + _OBJECTS.validate_python(llm_model_list), + aliases, + user_api_key_dict, + await _key_owner(user_api_key_dict), + llm_router, + ) + if aliases is not None + else _NOTHING_REACHABLE + ) + requested_model: Final = ( + _model_param_target( + model, + aliases, + _DEPLOYMENTS.validate_python(llm_model_list), + user_api_key_dict.team_id, + llm_router, + ) + if model is not None and aliases is not None + else model + ) _llm_model_list: Final = [ m for m in copy.deepcopy(llm_model_list) if not restrict_to_allowed_models - or _caller_may_probe_deployment(m, allowed_models, llm_router, user_api_key_dict.team_id, is_admin) + or _caller_may_probe_deployment(m, reachable, user_api_key_dict.team_id, is_admin) ] - targeted_ids: Final = _resolve_targeted_model_ids(_llm_model_list, model, model_id, user_api_key_dict.team_id) + targeted_ids: Final = _resolve_targeted_model_ids( + _llm_model_list, requested_model, model_id, user_api_key_dict.team_id + ) if restrict_to_allowed_models and targeted_ids is not None and not targeted_ids: raise HTTPException( status_code=status.HTTP_403_FORBIDDEN, @@ -1370,7 +1734,8 @@ async def health_endpoint( else: router_result: Final = await _perform_health_check_and_save( model_list=_llm_model_list, - target_model=target_model, + target_model=target_model if model_id else requested_model, + recorded_model_name=target_model, cli_model=None, details=health_check_details, prisma_client=prisma_client, diff --git a/tests/unit/proxy/auth/test_auth_checks_object_access_and_lookup.py b/tests/unit/proxy/auth/test_auth_checks_object_access_and_lookup.py index 0bdc5f1910a..6b59f6d40dc 100644 --- a/tests/unit/proxy/auth/test_auth_checks_object_access_and_lookup.py +++ b/tests/unit/proxy/auth/test_auth_checks_object_access_and_lookup.py @@ -47,6 +47,7 @@ from litellm.proxy.auth.auth_checks import ( can_object_call_model, _can_object_call_vector_stores, _check_agent_access_group_model_access, + agent_model_grants, check_end_user_budget, _check_team_member_budget, personal_key_owner_teams, @@ -11294,6 +11295,15 @@ async def test_agent_access_groups_naming_no_model_deny_every_model(): assert exc_info.value.type == ProxyErrorTypes.agent_model_access_denied +@pytest.mark.asyncio +async def test_agent_model_grants_names_every_model_the_agents_access_groups_allow(): + agent_key: Final = UserAPIKeyAuth(token="agent-token", agent_id="agent-1", models=[]) + resolve, asked = _agent_model_ceiling_resolver(frozenset({"gpt-5", "openai/gpt-5.6"})) + + assert await agent_model_grants(agent_key, resolve) == frozenset({"gpt-5", "openai/gpt-5.6"}) + assert asked == ["agent-1"] + + @pytest.mark.asyncio async def test_agent_without_access_groups_adds_no_model_ceiling(): agent_key: Final = UserAPIKeyAuth(token="agent-token", agent_id="agent-1", models=["gpt-5", "claude-sonnet"]) diff --git a/tests/unit/proxy/health_endpoints/test_health_endpoints.py b/tests/unit/proxy/health_endpoints/test_health_endpoints.py index 9847787efce..2258eb95fb8 100644 --- a/tests/unit/proxy/health_endpoints/test_health_endpoints.py +++ b/tests/unit/proxy/health_endpoints/test_health_endpoints.py @@ -1,8 +1,9 @@ import asyncio import copy +import itertools import json import time -from collections.abc import Awaitable, Iterator, Mapping, Sequence +from collections.abc import Awaitable, Callable, Iterator, Mapping, Sequence from contextlib import contextmanager from datetime import datetime, timedelta from importlib import import_module @@ -22,7 +23,7 @@ import litellm import litellm.proxy.health_endpoints._health_endpoints as _health_endpoints_module from litellm.litellm_core_utils.health_check_helpers import TEST_IMAGE_BASE64 from litellm.models.credentials import CredentialItem -from litellm.proxy._types import LitellmUserRoles, ProxyException, UserAPIKeyAuth +from litellm.proxy._types import LiteLLM_UserTable, LitellmUserRoles, ProxyErrorTypes, ProxyException, UserAPIKeyAuth from litellm.proxy.auth.user_api_key_auth import user_api_key_auth from litellm.proxy.health_endpoints._health_endpoints import ( _db_health_readiness_check, @@ -35,6 +36,7 @@ from litellm.proxy.health_endpoints._health_endpoints import ( test_model_connection as health_test_model_connection, ) from litellm.router import Router +from litellm.types.router import RouterModelGroupAliasItem from litellm.types.utils import oauth_token_exchange_litellm_params from litellm.types.workload_identity import ( ANTHROPIC_WIF_KWARGS_KEYS, @@ -925,7 +927,11 @@ async def test_test_model_connection_stored_audio_base_model_follows_the_stored_ deployment: Final = MappingProxyType( { "model_name": "custom-name-audio", - "litellm_params": {**deployment_params, "api_key": "fake-azure-key", "api_base": "https://x.openai.azure.com"}, + "litellm_params": { + **deployment_params, + "api_key": "fake-azure-key", + "api_base": "https://x.openai.azure.com", + }, "model_info": deployment_model_info, } ) @@ -1948,7 +1954,6 @@ def test_health_liveness_endpoint(proxy_client): assert duration_ms < 100, f"Health check took {duration_ms:.2f}ms, expected < 100ms for a simple endpoint" - def test_proxy_client_serves_full_proxy_routes_after_gateway_import(monkeypatch: pytest.MonkeyPatch) -> None: from litellm.proxy.proxy_server import app @@ -2472,6 +2477,7 @@ def _proxy_health_globals( llm_router: object, use_background_health_checks: bool = False, health_check_results: Mapping[str, object] | None = None, + prisma_client: object = None, ) -> Iterator[None]: with ( patch( # test-quality-ok: proxy module global, no injection seam @@ -2481,7 +2487,7 @@ def _proxy_health_globals( "litellm.proxy.proxy_server.llm_router", llm_router ), patch( # test-quality-ok: proxy module global, no injection seam - "litellm.proxy.proxy_server.prisma_client", None + "litellm.proxy.proxy_server.prisma_client", prisma_client ), patch( # test-quality-ok: proxy module global, no injection seam "litellm.proxy.proxy_server.use_background_health_checks", use_background_health_checks @@ -3812,7 +3818,11 @@ def test_clean_endpoint_data_never_displays_credential_fields(credential_field, async def _live_probed_model_ids( - model_list: Sequence[Mapping[str, object]], user_api_key_dict: UserAPIKeyAuth, model: str | None = None + model_list: Sequence[Mapping[str, object]], + user_api_key_dict: UserAPIKeyAuth, + model: str | None = None, + llm_router: Router | None = None, + prisma_client: object = None, ) -> set[str]: from fastapi import Response @@ -3825,7 +3835,7 @@ async def _live_probed_model_ids( return {"healthy_endpoints": [], "unhealthy_endpoints": [], "healthy_count": 0, "unhealthy_count": 0} with ( - _proxy_health_globals(model_list, _router_for(model_list)), + _proxy_health_globals(model_list, llm_router or _router_for(model_list), prisma_client=prisma_client), patch( # test-quality-ok: the model list handed to the probe is the assertion; no injection seam "litellm.proxy.health_endpoints._health_endpoints._perform_health_check_and_save", side_effect=fake_perform, @@ -4120,6 +4130,7 @@ async def _live_narrowed_model_ids( user_api_key_dict: UserAPIKeyAuth, model: str | None = None, model_id: str | None = None, + llm_router: Router | None = None, ) -> set[str]: from fastapi import Response @@ -4130,7 +4141,7 @@ async def _live_narrowed_model_ids( return probed, [], {} with ( - _proxy_health_globals(model_list, _router_for(model_list)), + _proxy_health_globals(model_list, llm_router or _router_for(model_list)), patch( # test-quality-ok: the probe is the provider edge; which deployments reach it is the assertion "litellm.proxy.health_check._perform_health_check", side_effect=fake_probe ), @@ -4289,6 +4300,699 @@ async def test_health_endpoint_keeps_a_team_only_public_name_off_a_team_less_key probe.assert_not_awaited() +_PROVIDER_PREFIXED_MODEL_LIST: Final = [ + { + "model_name": "openai/gpt-5.4-nano", + "litellm_params": {"model": "openai/gpt-5.4-nano"}, + "model_info": {"id": "id-prefixed-nano"}, + }, + { + "model_name": "gpt-5.4-mini", + "litellm_params": {"model": "openai/gpt-5.4-mini"}, + "model_info": {"id": "id-mini"}, + }, + { + "model_name": "claude-haiku-5-5", + "litellm_params": {"model": "anthropic/claude-haiku-5-5"}, + "model_info": {"id": "id-haiku"}, + }, +] +_PROVIDER_PREFIXED_CACHED_RESULTS: Final = { + "healthy_endpoints": [ + {"model": "openai/gpt-5.4-nano", "model_id": "id-prefixed-nano"}, + {"model": "openai/gpt-5.4-mini", "model_id": "id-mini"}, + {"model": "anthropic/claude-haiku-5-5", "model_id": "id-haiku"}, + ], + "unhealthy_endpoints": [], + "healthy_count": 3, + "unhealthy_count": 0, +} +_PROVIDER_WILDCARD_KEY: Final = UserAPIKeyAuth(api_key="hashed-test-key", models=["openai/*"]) +_RESTRICTED_TEAM_KEY: Final = UserAPIKeyAuth( + api_key="hashed-test-key", models=[], team_id="team-a", team_models=["gpt-5.4-mini"] +) + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + ("caller", "expected_ids"), + [(_PROVIDER_WILDCARD_KEY, {"id-prefixed-nano", "id-mini"}), (_RESTRICTED_TEAM_KEY, {"id-mini"})], +) +async def test_health_endpoint_probes_the_deployments_auth_lets_the_key_request( + caller: UserAPIKeyAuth, expected_ids: set[str] +): + probed: Final = await _live_probed_model_ids(_PROVIDER_PREFIXED_MODEL_LIST, caller) + + assert probed == expected_ids + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + ("caller", "expected_ids"), + [(_PROVIDER_WILDCARD_KEY, ["id-prefixed-nano", "id-mini"]), (_RESTRICTED_TEAM_KEY, ["id-mini"])], +) +async def test_health_endpoint_returns_the_deployments_auth_lets_the_key_request_on_background_cache_path( + caller: UserAPIKeyAuth, expected_ids: list[str] +): + from fastapi import Response + + from litellm.proxy.health_endpoints._health_endpoints import health_endpoint + + with _proxy_health_globals( + _PROVIDER_PREFIXED_MODEL_LIST, + _router_for(_PROVIDER_PREFIXED_MODEL_LIST), + use_background_health_checks=True, + health_check_results=_PROVIDER_PREFIXED_CACHED_RESULTS, + ): + result: Final = await health_endpoint(response=Response(), user_api_key_dict=caller, model=None, model_id=None) + + assert [ep["model_id"] for ep in result["healthy_endpoints"]] == expected_ids + + +@pytest.mark.asyncio +@pytest.mark.parametrize("use_background_health_checks", [False, True]) +async def test_health_endpoint_refuses_a_deployment_the_team_allowlist_forbids(use_background_health_checks: bool): + from fastapi import HTTPException, Response + + from litellm.proxy.health_endpoints._health_endpoints import health_endpoint + + with ( + _proxy_health_globals( + _PROVIDER_PREFIXED_MODEL_LIST, + _router_for(_PROVIDER_PREFIXED_MODEL_LIST), + use_background_health_checks=use_background_health_checks, + health_check_results=_PROVIDER_PREFIXED_CACHED_RESULTS, + ), + patch( # test-quality-ok: the probe must never run; the endpoint has no injection seam for it + "litellm.proxy.health_endpoints._health_endpoints._perform_health_check_and_save", new_callable=AsyncMock + ) as probe, + pytest.raises(HTTPException) as refused, + ): + await health_endpoint( + response=Response(), user_api_key_dict=_RESTRICTED_TEAM_KEY, model=None, model_id="id-prefixed-nano" + ) + + assert refused.value.status_code == 403 + probe.assert_not_awaited() + + +_KEY_ALIASED_AWAY_FROM_HAIKU: Final = UserAPIKeyAuth( + api_key="hashed-test-key", models=["gpt-5.4-mini"], aliases={"claude-haiku-5-5": "gpt-5.4-mini"} +) +_TEAM_ALIASED_AWAY_FROM_HAIKU: Final = UserAPIKeyAuth( + api_key="hashed-test-key", + models=["claude-haiku-5-5"], + team_id="team-a", + team_model_aliases={"claude-haiku-5-5": "gpt-5.4-mini"}, +) +_MINI_ONLY_KEY: Final = UserAPIKeyAuth(api_key="hashed-test-key", models=["gpt-5.4-mini"]) +_ALIAS_SHADOW_CASES: Final = [ + pytest.param(_KEY_ALIASED_AWAY_FROM_HAIKU, {}, id="key-alias"), + pytest.param(_TEAM_ALIASED_AWAY_FROM_HAIKU, {}, id="team-alias"), + pytest.param(_MINI_ONLY_KEY, {"claude-haiku-5-5": "gpt-5.4-mini"}, id="router-alias"), +] + + +def _router_with_aliases(model_group_alias: Mapping[str, str]) -> Router: + return Router(model_list=copy.deepcopy(_PROVIDER_PREFIXED_MODEL_LIST), model_group_alias=dict(model_group_alias)) + + +@pytest.mark.asyncio +@pytest.mark.parametrize(("caller", "model_group_alias"), _ALIAS_SHADOW_CASES) +async def test_health_endpoint_probes_the_deployment_an_alias_sends_the_key_to_not_the_one_it_shadows( + caller: UserAPIKeyAuth, model_group_alias: Mapping[str, str] +): + probed: Final = await _live_probed_model_ids( + _PROVIDER_PREFIXED_MODEL_LIST, caller, llm_router=_router_with_aliases(model_group_alias) + ) + + assert probed == {"id-mini"} + + +@pytest.mark.asyncio +@pytest.mark.parametrize(("caller", "model_group_alias"), _ALIAS_SHADOW_CASES) +async def test_health_endpoint_returns_the_deployment_an_alias_sends_the_key_to_on_background_cache_path( + caller: UserAPIKeyAuth, model_group_alias: Mapping[str, str] +): + from fastapi import Response + + from litellm.proxy.health_endpoints._health_endpoints import health_endpoint + + with _proxy_health_globals( + _PROVIDER_PREFIXED_MODEL_LIST, + _router_with_aliases(model_group_alias), + use_background_health_checks=True, + health_check_results=_PROVIDER_PREFIXED_CACHED_RESULTS, + ): + result: Final = await health_endpoint(response=Response(), user_api_key_dict=caller, model=None, model_id=None) + + assert [ep["model_id"] for ep in result["healthy_endpoints"]] == ["id-mini"] + + +@pytest.mark.asyncio +@pytest.mark.parametrize("use_background_health_checks", [False, True]) +async def test_health_endpoint_refuses_the_deployment_a_key_alias_shadows(use_background_health_checks: bool): + from fastapi import HTTPException, Response + + from litellm.proxy.health_endpoints._health_endpoints import health_endpoint + + with ( + _proxy_health_globals( + _PROVIDER_PREFIXED_MODEL_LIST, + _router_for(_PROVIDER_PREFIXED_MODEL_LIST), + use_background_health_checks=use_background_health_checks, + health_check_results=_PROVIDER_PREFIXED_CACHED_RESULTS, + ), + patch( # test-quality-ok: the probe must never run; the endpoint has no injection seam for it + "litellm.proxy.health_endpoints._health_endpoints._perform_health_check_and_save", new_callable=AsyncMock + ) as probe, + pytest.raises(HTTPException) as refused, + ): + await health_endpoint( + response=Response(), user_api_key_dict=_KEY_ALIASED_AWAY_FROM_HAIKU, model=None, model_id="id-haiku" + ) + + assert refused.value.status_code == 403 + probe.assert_not_awaited() + + +@pytest.mark.asyncio +@pytest.mark.parametrize(("caller", "model_group_alias"), _ALIAS_SHADOW_CASES) +async def test_health_endpoint_targets_the_deployment_an_alias_name_sends_the_key_to( + caller: UserAPIKeyAuth, model_group_alias: Mapping[str, str] +): + probed: Final = await _live_narrowed_model_ids( + _PROVIDER_PREFIXED_MODEL_LIST, + caller, + model="claude-haiku-5-5", + llm_router=_router_with_aliases(model_group_alias), + ) + + assert probed == {"id-mini"} + + +@pytest.mark.asyncio +@pytest.mark.parametrize(("caller", "model_group_alias"), _ALIAS_SHADOW_CASES) +async def test_health_endpoint_records_an_alias_check_under_the_name_the_caller_asked_for( + caller: UserAPIKeyAuth, model_group_alias: Mapping[str, str] +): + from fastapi import Response + + from litellm.proxy.health_endpoints._health_endpoints import health_endpoint + + async def fake_probe(model_list, details=True, max_concurrency=None, instrumentation_context=None): + return [{"model": m["litellm_params"]["model"], "model_id": m["model_info"]["id"]} for m in model_list], [], {} + + with ( + _proxy_health_globals( + _PROVIDER_PREFIXED_MODEL_LIST, _router_with_aliases(model_group_alias), prisma_client=MagicMock() + ), + patch( # test-quality-ok: the probe is the provider edge; the endpoint has no injection seam for it + "litellm.proxy.health_check._perform_health_check", side_effect=fake_probe + ), + patch( # test-quality-ok: the history row is the assertion; the endpoint has no injection seam for its writer + "litellm.proxy.health_endpoints._health_endpoints._save_health_check_to_db", new_callable=AsyncMock + ) as save, + ): + result: Final = await health_endpoint( + response=Response(), user_api_key_dict=caller, model="claude-haiku-5-5", model_id=None + ) + + assert [ep["model_id"] for ep in result["healthy_endpoints"]] == ["id-mini"] + assert save.call_args.args[1] == "claude-haiku-5-5" + + +@pytest.mark.asyncio +@pytest.mark.parametrize(("caller", "model_group_alias"), _ALIAS_SHADOW_CASES) +async def test_health_endpoint_targets_the_deployment_an_alias_name_sends_the_key_to_on_background_cache_path( + caller: UserAPIKeyAuth, model_group_alias: Mapping[str, str] +): + from fastapi import Response + + from litellm.proxy.health_endpoints._health_endpoints import health_endpoint + + with _proxy_health_globals( + _PROVIDER_PREFIXED_MODEL_LIST, + _router_with_aliases(model_group_alias), + use_background_health_checks=True, + health_check_results=_PROVIDER_PREFIXED_CACHED_RESULTS, + ): + result: Final = await health_endpoint( + response=Response(), user_api_key_dict=caller, model="claude-haiku-5-5", model_id=None + ) + + assert [ep["model_id"] for ep in result["healthy_endpoints"]] == ["id-mini"] + + +@pytest.mark.asyncio +async def test_health_endpoint_never_probes_a_deployment_without_a_model_name_for_a_restricted_key(): + nameless_deployment: Final = { + "litellm_params": {"model": "openai/gpt-5.4-mini"}, + "model_info": {"id": "id-nameless"}, + } + + probed: Final = await _live_probed_model_ids( + [*_PROVIDER_PREFIXED_MODEL_LIST, nameless_deployment], + _MINI_ONLY_KEY, + llm_router=_router_for(_PROVIDER_PREFIXED_MODEL_LIST), + ) + + assert probed == {"id-mini"} + + +@pytest.mark.asyncio +async def test_health_endpoint_never_probes_the_deployment_a_keys_router_alias_sends_it_to_but_forbids(): + caller: Final = UserAPIKeyAuth( + api_key="hashed-test-key", + models=["gpt-5.4-mini", "openai/gpt-5.4-nano"], + router_settings={"model_group_alias": {"gpt-5.4-mini": "claude-haiku-5-5"}}, + ) + + probed: Final = await _live_probed_model_ids(_PROVIDER_PREFIXED_MODEL_LIST, caller) + + assert probed == {"id-prefixed-nano"} + + +_WILDCARD_DEPLOYMENT: Final = { + "model_name": "openai/*", + "litellm_params": {"model": "openai/*"}, + "model_info": {"id": "id-wild"}, +} +_CATCH_ALL_DEPLOYMENT: Final = {"model_name": "*", "litellm_params": {"model": "*"}, "model_info": {"id": "id-star"}} +_TEAM_WILDCARD_DEPLOYMENT: Final = { + "model_name": "openai-wild_team-a_7c1e", + "litellm_params": {"model": "openai/*"}, + "model_info": {"id": "id-team-wild", "team_id": "team-a", "team_public_model_name": "openai/*"}, +} + + +def _team_key(team_models: list[str], team_model_aliases: Mapping[str, str] | None = None) -> UserAPIKeyAuth: + return UserAPIKeyAuth( + api_key="hashed-test-key", + models=[], + team_id="team-a", + team_models=team_models, + team_model_aliases=dict(team_model_aliases) if team_model_aliases is not None else None, + ) + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + ("model_list", "model_group_alias", "caller", "expected_ids"), + [ + pytest.param( + [_WILDCARD_DEPLOYMENT, _PROVIDER_PREFIXED_MODEL_LIST[2]], + {}, + _team_key(["openai/gpt-5.4-nano"]), + {"id-wild"}, + id="team-name-served-by-provider-wildcard", + ), + pytest.param( + [_TEAM_WILDCARD_DEPLOYMENT, _PROVIDER_PREFIXED_MODEL_LIST[2]], + {}, + _team_key(["openai/gpt-5.4-nano"]), + {"id-team-wild"}, + id="team-name-served-by-teams-own-wildcard", + ), + pytest.param( + [_CATCH_ALL_DEPLOYMENT], + {}, + _team_key(["gpt-5.4-mini"]), + {"id-star"}, + id="team-name-served-by-catch-all", + ), + pytest.param( + _PROVIDER_PREFIXED_MODEL_LIST, + {"gpt-4": "gpt-5.4-mini"}, + _team_key(["gpt-4"]), + {"id-mini"}, + id="team-allows-router-alias", + ), + pytest.param( + _PROVIDER_PREFIXED_MODEL_LIST, + {}, + _team_key(["my-alias"], {"my-alias": "gpt-5.4-mini"}), + {"id-mini"}, + id="team-allows-team-alias", + ), + pytest.param( + _PROVIDER_PREFIXED_MODEL_LIST, + {"friendly-mini": RouterModelGroupAliasItem(model="gpt-5.4-mini", hidden=True)}, + UserAPIKeyAuth(api_key="hashed-test-key", models=["friendly-*"]), + {"id-mini"}, + id="wildcard-key-allows-hidden-router-alias", + ), + pytest.param( + _PROVIDER_PREFIXED_MODEL_LIST, + {}, + _team_key(["friendly-*"], {"friendly-mini": "gpt-5.4-mini"}), + {"id-mini"}, + id="wildcard-team-allows-team-alias", + ), + pytest.param( + _PROVIDER_PREFIXED_MODEL_LIST, + {"claude-haiku-5-5": "gpt-5.4-mini"}, + UserAPIKeyAuth(api_key="hashed-test-key", models=[]), + {"id-prefixed-nano", "id-mini", "id-haiku"}, + id="unrestricted-key-calls-shadowed-deployment-by-id", + ), + ], +) +async def test_health_endpoint_probes_every_deployment_the_callers_allowed_names_route_to( + model_list: Sequence[Mapping[str, object]], + model_group_alias: Mapping[str, str | RouterModelGroupAliasItem], + caller: UserAPIKeyAuth, + expected_ids: set[str], +): + llm_router: Final = Router( + model_list=copy.deepcopy([dict(deployment) for deployment in model_list]), + model_group_alias=dict(model_group_alias), + ) + + probed: Final = await _live_probed_model_ids(llm_router.get_model_list() or [], caller, llm_router=llm_router) + + assert probed == expected_ids + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "caller", + [ + pytest.param( + UserAPIKeyAuth(api_key="hashed-test-key", models=["gpt-5.4-mini", "gpt-4-retired"]), id="key-grant" + ), + pytest.param(_team_key(["gpt-5.4-mini", "gpt-4-retired"]), id="team-grant"), + ], +) +async def test_health_endpoint_never_probes_the_default_fallback_for_a_granted_name_no_deployment_serves( + caller: UserAPIKeyAuth, +) -> None: + llm_router: Final = Router( + model_list=copy.deepcopy(_PROVIDER_PREFIXED_MODEL_LIST), default_fallbacks=["claude-haiku-5-5"] + ) + + probed: Final = await _live_probed_model_ids(_PROVIDER_PREFIXED_MODEL_LIST, caller, llm_router=llm_router) + + assert probed == {"id-mini"} + + +def _personal_key_whose_owner_may_call(models: list[str]) -> UserAPIKeyAuth: + from litellm.proxy.proxy_server import user_api_key_cache + + owner_id: Final = "health-scope-owner-" + "-".join(models) + user_api_key_cache.set_cache(key=owner_id, value=LiteLLM_UserTable(user_id=owner_id, models=models)) + return UserAPIKeyAuth(api_key="hashed-test-key", models=[], user_id=owner_id) + + +def _managed_agent_key_granted(models: list[str]) -> UserAPIKeyAuth: + from litellm.types.agents import AgentResponse + + policy: Final = AgentResponse( + agent_id="managed", agent_name="Managed", agent_card_params={}, object_permission={"models": models} + ) + return UserAPIKeyAuth(api_key="hashed-test-key", models=[]).model_copy(update={"managed_agent_policy": policy}) + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + ("make_caller", "models", "expected_ids"), + [ + pytest.param(_personal_key_whose_owner_may_call, ["gpt-5.4-mini"], {"id-mini"}, id="key-owners-models"), + pytest.param(_managed_agent_key_granted, ["gpt-5.4-mini"], {"id-mini"}, id="managed-agents-models"), + pytest.param(_managed_agent_key_granted, [], set(), id="managed-agent-without-model-grants"), + ], +) +async def test_health_endpoint_probes_only_the_models_the_keys_user_or_agent_may_call( + make_caller: Callable[[list[str]], UserAPIKeyAuth], models: list[str], expected_ids: set[str] +): + probed: Final = await _live_probed_model_ids(_PROVIDER_PREFIXED_MODEL_LIST, make_caller(models)) + + assert probed == expected_ids + + +@pytest.mark.asyncio +async def test_health_endpoint_refuses_a_model_id_the_keys_owner_may_not_call(): + from fastapi import Response + + from litellm.proxy.health_endpoints._health_endpoints import health_endpoint + + with ( + _proxy_health_globals(_PROVIDER_PREFIXED_MODEL_LIST, _router_for(_PROVIDER_PREFIXED_MODEL_LIST)), + patch( # test-quality-ok: the probe must never run; the endpoint has no injection seam for it + "litellm.proxy.health_endpoints._health_endpoints._perform_health_check_and_save", new_callable=AsyncMock + ) as probe, + pytest.raises(HTTPException) as refused, + ): + await health_endpoint( + response=Response(), + user_api_key_dict=_personal_key_whose_owner_may_call(["gpt-5.4-mini"]), + model=None, + model_id="id-prefixed-nano", + ) + + assert refused.value.status_code == 403 + probe.assert_not_awaited() + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "make_caller", + [ + pytest.param(_personal_key_whose_owner_may_call, id="key-owner"), + pytest.param(_managed_agent_key_granted, id="managed-agent"), + ], +) +async def test_health_endpoint_keeps_the_wildcard_serving_a_name_the_keys_user_or_agent_may_call( + make_caller: Callable[[list[str]], UserAPIKeyAuth], +): + model_list: Final = [_WILDCARD_DEPLOYMENT, *_PROVIDER_PREFIXED_MODEL_LIST] + + probed: Final = await _live_probed_model_ids(model_list, make_caller(["openai/gpt-5.6"])) + + assert probed == {"id-wild"} + + +@pytest.mark.asyncio +async def test_health_endpoint_answers_an_agent_key_acting_for_an_unknown_user_the_way_chat_does(): + from litellm.types.agents import AgentCaller + + users_table: Final = MagicMock() + users_table.db.litellm_usertable.find_unique = AsyncMock(return_value=None) + agent_key: Final = UserAPIKeyAuth(api_key="hashed-test-key", models=["gpt-5.4-mini"]).model_copy( + update={"agent_caller": AgentCaller(user_id="user-deleted-before-this-request")} + ) + + with pytest.raises(ProxyException) as failed: + await _live_probed_model_ids(_PROVIDER_PREFIXED_MODEL_LIST, agent_key, prisma_client=users_table) + + assert (failed.value.code, failed.value.type) == ("401", ProxyErrorTypes.auth_error) + assert "user-deleted-before-this-request" in failed.value.message + + +@pytest.mark.asyncio +async def test_health_endpoint_lets_other_requests_run_while_it_scopes_a_large_model_list(): + model_list: Final = [ + {"model_name": f"model-{n}", "litellm_params": {"model": f"openai/model-{n}"}, "model_info": {"id": f"id-{n}"}} + for n in range(50) + ] + other_request_turns: Final = itertools.count() + scoping_done: Final = asyncio.Event() + + async def scope() -> set[str]: + try: + return await _live_probed_model_ids( + model_list, UserAPIKeyAuth(api_key="hashed-test-key", models=["model-0"]) + ) + finally: + scoping_done.set() + + async def other_request() -> None: + while not scoping_done.is_set(): + next(other_request_turns) + await asyncio.sleep(0) + + probed, _ = await asyncio.gather(scope(), other_request()) + + assert probed == {"id-0"} + assert next(other_request_turns) >= len(model_list) + + +_WILDCARD_CACHED_RESULTS: Final = { + "healthy_endpoints": [ + {"model": "openai/gpt-5.6", "model_id": "id-wild"}, + {"model": "openai/gpt-5.6", "model_id": "id-team-wild"}, + {"model": "anthropic/claude-haiku-5-5", "model_id": "id-haiku"}, + ], + "unhealthy_endpoints": [], + "healthy_count": 3, + "unhealthy_count": 0, +} + + +@pytest.mark.asyncio +@pytest.mark.parametrize("use_background_health_checks", [False, True]) +@pytest.mark.parametrize( + ("wildcard_deployment", "expected_id"), + [ + pytest.param(_WILDCARD_DEPLOYMENT, "id-wild", id="proxy-wildcard"), + pytest.param(_TEAM_WILDCARD_DEPLOYMENT, "id-team-wild", id="teams-own-wildcard"), + ], +) +async def test_health_endpoint_model_param_reaches_the_wildcard_serving_a_team_allowed_name( + wildcard_deployment: Mapping[str, object], expected_id: str, use_background_health_checks: bool +): + from fastapi import Response + + from litellm.proxy.health_endpoints._health_endpoints import health_endpoint + + model_list: Final = [wildcard_deployment, _PROVIDER_PREFIXED_MODEL_LIST[2]] + + async def fake_probe(model_list, details=True, max_concurrency=None, instrumentation_context=None): + return [{"model": m["litellm_params"]["model"], "model_id": m["model_info"]["id"]} for m in model_list], [], {} + + with ( + _proxy_health_globals( + model_list, + _router_for(model_list), + use_background_health_checks=use_background_health_checks, + health_check_results=_WILDCARD_CACHED_RESULTS, + ), + patch( # test-quality-ok: the probe is the provider edge; which deployments reach it is the assertion + "litellm.proxy.health_check._perform_health_check", side_effect=fake_probe + ), + ): + result: Final = await health_endpoint( + response=Response(), user_api_key_dict=_team_key(["openai/gpt-5.6"]), model="openai/gpt-5.6", model_id=None + ) + + assert [ep["model_id"] for ep in result["healthy_endpoints"]] == [expected_id] + + +_ACCESS_GROUP_SPLIT_MODEL_LIST: Final = [ + { + "model_name": "gpt-5.4-mini", + "litellm_params": {"model": "openai/gpt-5.4-mini"}, + "model_info": {"id": "id-mini-premium", "access_groups": ["premium"]}, + }, + { + "model_name": "gpt-5.4-mini", + "litellm_params": {"model": "openai/gpt-5.4-mini"}, + "model_info": {"id": "id-mini-standard"}, + }, +] + + +@pytest.mark.asyncio +async def test_health_endpoint_probes_only_the_access_groups_deployments_of_a_shared_name() -> None: + caller: Final = UserAPIKeyAuth(api_key="hashed-test-key", models=["premium"]) + + probed: Final = await _live_probed_model_ids(_ACCESS_GROUP_SPLIT_MODEL_LIST, caller) + + assert probed == {"id-mini-premium"} + + +@pytest.mark.asyncio +@pytest.mark.parametrize("direct_grant", ["gpt-5.4-mini", "*", "all-proxy-models"]) +async def test_health_endpoint_probes_every_deployment_of_a_name_the_key_also_holds_directly(direct_grant: str) -> None: + caller: Final = UserAPIKeyAuth(api_key="hashed-test-key", models=["premium", direct_grant]) + + probed: Final = await _live_probed_model_ids(_ACCESS_GROUP_SPLIT_MODEL_LIST, caller) + + assert probed == {"id-mini-premium", "id-mini-standard"} + + +_ACCESS_GROUP_SPLIT_WILDCARD_MODEL_LIST: Final = [ + { + "model_name": "openai/*", + "litellm_params": {"model": "openai/*"}, + "model_info": {"id": "id-wild-premium", "access_groups": ["premium"]}, + }, + { + "model_name": "openai/*", + "litellm_params": {"model": "openai/*"}, + "model_info": {"id": "id-wild-standard", "access_groups": ["standard"]}, + }, +] + + +@pytest.mark.asyncio +async def test_health_endpoint_probes_only_the_access_groups_wildcard_deployments_serving_a_name() -> None: + caller: Final = _personal_key_whose_owner_may_call(["openai/gpt-5.4-nano"]).model_copy( + update={"models": ["premium"]} + ) + + probed: Final = await _live_probed_model_ids(_ACCESS_GROUP_SPLIT_WILDCARD_MODEL_LIST, caller) + + assert probed == {"id-wild-premium"} + + +def _team_deployment( + internal_name: str, public_name: str, model: str, model_id: str, access_groups: list[str] +) -> Mapping[str, object]: + return { + "model_name": internal_name, + "litellm_params": {"model": model}, + "model_info": { + "id": model_id, + "team_id": "team-a", + "team_public_model_name": public_name, + "access_groups": access_groups, + }, + } + + +def _team_split_model_list(public_name: str, model: str) -> list[Mapping[str, object]]: + return [ + _team_deployment("model_name_team-a_7c1e4b2a", public_name, model, "id-team-premium", ["premium"]), + _team_deployment("model_name_team-a_9d2f6e01", public_name, model, "id-team-standard", ["standard"]), + ] + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + ("public_name", "model"), + [ + pytest.param("openai/*", "openai/*", id="team-wildcard"), + pytest.param("gpt-5.4-mini", "openai/gpt-5.4-mini", id="team-model"), + ], +) +async def test_health_endpoint_probes_only_the_access_groups_team_deployments_serving_a_name( + public_name: str, model: str +) -> None: + probed: Final = await _live_probed_model_ids(_team_split_model_list(public_name, model), _team_key(["premium"])) + + assert probed == {"id-team-premium"} + + +@pytest.mark.asyncio +@pytest.mark.parametrize("internal_name_grant", ["model_name_team-a_9d2f6e01", "model_name_team-a_*"]) +async def test_health_endpoint_probes_a_team_deployment_whose_internal_name_the_key_holds( + internal_name_grant: str, +) -> None: + caller: Final = _team_key(["premium", internal_name_grant]) + + probed: Final = await _live_probed_model_ids(_team_split_model_list("gpt-5.4-mini", "openai/gpt-5.4-mini"), caller) + + assert probed == {"id-team-premium", "id-team-standard"} + + +@pytest.mark.asyncio +async def test_health_endpoint_probes_only_the_team_deployment_whose_internal_name_an_agent_holds() -> None: + from litellm.types.agents import AgentResponse + + policy: Final = AgentResponse( + agent_id="managed", + agent_name="Managed", + agent_card_params={}, + object_permission={"models": ["model_name_team-a_9d2f6e01"]}, + ) + caller: Final = _team_key(["gpt-5.4-mini"]).model_copy(update={"managed_agent_policy": policy}) + + probed: Final = await _live_probed_model_ids(_team_split_model_list("gpt-5.4-mini", "openai/gpt-5.4-mini"), caller) + + assert probed == {"id-team-standard"} + + def test_resolve_targeted_model_ids_lets_model_id_win_over_model(): resolve = _health_endpoints_module._resolve_targeted_model_ids diff --git a/tests/unit/proxy/test_health_check_functions.py b/tests/unit/proxy/test_health_check_functions.py index 2c0198cd538..79eec44958f 100644 --- a/tests/unit/proxy/test_health_check_functions.py +++ b/tests/unit/proxy/test_health_check_functions.py @@ -741,6 +741,7 @@ async def test_perform_health_check_and_save_passes_model_id_to_perform_health_c result = await _perform_health_check_and_save( model_list=model_list, target_model=None, + recorded_model_name=None, cli_model=None, details=True, prisma_client=None, @@ -777,6 +778,7 @@ async def test_perform_health_check_and_save_forwards_skip_disabled_background_f await _perform_health_check_and_save( model_list=model_list, target_model=None, + recorded_model_name=None, cli_model=None, details=True, prisma_client=None, From 43c07a0f1246c95da710408df4d7846bc5f848e2 Mon Sep 17 00:00:00 2001 From: tavren27 <348969968+tavren27@users.noreply.github.com> Date: Fri, 30 Oct 2026 07:48:32 -0700 Subject: [PATCH 05/18] fix(router)!: check and count deployment itpm and otpm on every Responses WebSocket turn (#49069) * fix(router): check and count deployment itpm and otpm on every Responses WebSocket turn The success hook only reconciled an ITPM/OTPM reservation found in top-level metadata or the standard logging object, but on the proxy the reservation lives in litellm_params.metadata, so no reservation was ever settled there. The hook now always calls the reconcile, which already returns early without a reservation and clears what it settled, so a second success event is a no-op. A Responses WebSocket turn after the first now has its deployment's current-minute ITPM and OTPM counters checked and is refused with a 429 frame once a counter is at its limit. An admitted turn carries a zero reservation on the current minute's keys, so its actual tokens land on those counters on both the native and the managed relay. * fix(responses): settle the socket open's itpm/otpm reservation on the first turn only After the first turn takes its logging copy, both websocket relays drop the open's ITPM/OTPM reservation keys from the connection metadata, the same way the budget reservation is detached. Later turns, unsolicited backend events, turns after limits are removed mid-socket, and sockets with no turn admission no longer settle or refund the open's reservation a second time. * refactor(responses): read the itpm/otpm reservation keys from core metadata helpers instead of router_utils * fix(responses): keep a socket open's hold for its first real turn and check tokens before request limits A turn that will settle the socket open's itpm/otpm reservation no longer counts that reservation against itself or swaps it for a zero one, so Codex's warmup frame on a managed provider, or a refused first frame on the native socket, no longer blocks the first real prompt or strands the hold for the minute. The token check now runs before the rpm/rpd checks, so a turn refused for tokens never uses up a request slot. Token-check tests pin the clock, and the managed relay settle test drives a real streamed call through a custom provider. * test(router): pin the router clock in the socket turn admission tests Both tests seeded this minute and the next to dodge a minute boundary. They now use the file's pinned-minute fixture and seed only that minute, so neither depends on when the test runs. * fix(router): check a responses socket's open without reserving and read shared token counters first A Responses WebSocket open now runs the same itpm/otpm check a turn runs and reserves nothing. It used to reserve an estimate that falls back to the model's max output, so a socket could not open under any otpm below that. The open still records a zero reservation for its own minute, so the first frame, which the relay forwards without a turn check, counts its actual tokens like every later turn. This drops the plumbing that only carried the open's hold into the first turn. The turn admission callable takes no arguments again, and the native relay detaches only the budget reservation, as before this change. The itpm/otpm turn check reads Redis first and falls back to this instance's in-memory count only without Redis or when the circuit breaker is open. These counters go down when a call settles, so an in-memory copy at the limit could refuse turns after the shared count dropped. The tpm check keeps its in-memory shortcut, since its counter only grows within a minute. Responses usage that reports input_tokens_details.cached_tokens now leaves cached input out of itpm, as chat usage already did. * fix(router): leave a socket turn's cached input out of itpm The native Responses WebSocket relay logs a turn with its list of events, so the deployment rate limit hook finds no usage on the response and falls back to the standard logging payload. That fallback read cached input from metadata.cache_read_input_tokens, a key the payload's cleaned metadata never carries, so every cached token of a socket turn was counted against itpm. The fallback now reads the cached count from the payload's usage_object, the same usage the logging path already built for cost, so a native socket turn counts only its uncached input. The managed relay and HTTP /v1/responses already counted correctly through the response's own usage * fix(router): mark only responses socket opens as check-only and count a turn on the minute it finishes The check-only branch for a socket open keyed on a websocket value in the request kwargs, which any HTTP body can carry and which the realtime route also sets. The router now passes an opens_responses_socket flag into set_io_token_rate_limit_request_kwargs only for its _aresponses_websocket call, and the flag lives in the same context value as the request kwargs, so no request body can set it and realtime opens reserve and get refused as before A zero reservation holds nothing to refund, so a socket turn now counts its real tokens on the minute the call finishes, read from the clock at settle time. A non-zero reservation still settles on the keys it reserved on so its refund hits the same counter When the shared Redis read returns nothing, the check now falls back to this instance's in-memory count instead of admitting the turn with no count at all * fix(router): keep io token reservation keys out of the metadata sent to providers The itpm/otpm pre-call check wrote its reservation keys into both metadata and litellm_metadata. On the routes the router serves through its generic helper (Responses, Anthropic Messages, and the rest), metadata is the client's own field and goes to the provider as is. So _litellm_itpm_cache_key and _litellm_otpm_cache_key, which carry the deployment id and the model name, reached the provider, came back in the echoed response metadata, and counted toward the provider's 16-key cap. The managed Responses socket relay merged each turn's reservation into the frame's metadata the same way, and a native socket opened with metadata sent the keys on every turn through its request defaults The router now passes the limiter the metadata field it uses for its own fields on that call (metadata for chat, litellm_metadata for the generic routes), and the reservation goes only there. metadata_channel is a required keyword, so a new caller has to name the field it writes. The managed relay puts each turn's reservation only on litellm_metadata With the reservation only on litellm_metadata, a retried call on the generic routes refunded its first attempt twice and pushed the deployment's counters below zero. The failure hook read the reservation from the logging copy in litellm_params.metadata and cleared only that copy, so the router's litellm_metadata still held it and the retry refunded it again. The hooks now also read and clear litellm_params.litellm_metadata, which is the router's own dict The proxy does not strip client keys that start with _litellm_ on the way in, so dropping that prefix at the provider boundary would also drop keys clients send today The io token tests now run Redis in process with fakeredis instead of a TCP fake server * test(responses): count socket turns through real logging callbacks instead of a Logging subclass The native relay test now hands the rate limit check to a real Logging object as its success and failure callback, so the turn's tokens go through the same dispatch path production uses. --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> --- .../internal_call_metadata.py | 6 + litellm/llms/custom_httpx/llm_http_handler.py | 4 +- litellm/responses/main.py | 6 +- litellm/responses/streaming_iterator.py | 103 ++-- litellm/router.py | 26 +- .../io_token_rate_limit_check.py | 209 ++++++--- ...test_responses_websocket_turn_admission.py | 113 ++++- .../test_responses_websocket_all_providers.py | 441 +++++++++++++++++- .../test_router/test_io_token_rate_limits.py | 356 ++++++++++++-- tests/unit/test_router/test_router.py | 6 +- .../test_router/test_router_helper_utils.py | 374 ++++++++++++++- 11 files changed, 1492 insertions(+), 152 deletions(-) diff --git a/litellm/litellm_core_utils/internal_call_metadata.py b/litellm/litellm_core_utils/internal_call_metadata.py index f3dbd612bee..453d4358b35 100644 --- a/litellm/litellm_core_utils/internal_call_metadata.py +++ b/litellm/litellm_core_utils/internal_call_metadata.py @@ -35,6 +35,12 @@ from litellm.types.utils import ( BUDGET_RESERVATION_METADATA_KEYS: Final = frozenset({"user_api_key_budget_reservation"}) +ITPM_RESERVED_KEY: Final = "_litellm_itpm_reserved" +OTPM_RESERVED_KEY: Final = "_litellm_otpm_reserved" +ITPM_CACHE_KEY: Final = "_litellm_itpm_cache_key" +OTPM_CACHE_KEY: Final = "_litellm_otpm_cache_key" +IO_TOKEN_RESERVATION_KEYS: Final = frozenset({ITPM_RESERVED_KEY, OTPM_RESERVED_KEY, ITPM_CACHE_KEY, OTPM_CACHE_KEY}) + MODEL_ACCESS_GROUP_METADATA_KEY: Final = "user_api_key_matched_model_access_groups" """Where auth records the model access groups that authorized the request, for the spend writer. diff --git a/litellm/llms/custom_httpx/llm_http_handler.py b/litellm/llms/custom_httpx/llm_http_handler.py index 61fc23589ac..9a2b352dd86 100644 --- a/litellm/llms/custom_httpx/llm_http_handler.py +++ b/litellm/llms/custom_httpx/llm_http_handler.py @@ -249,7 +249,7 @@ if TYPE_CHECKING: FakeAnthropicMessagesStreamIterator, ) from litellm.llms.base_llm.passthrough.transformation import BasePassthroughConfig - from litellm.responses.streaming_iterator import ResponsesWsFrameGuard + from litellm.responses.streaming_iterator import ResponsesWsFrameGuard, ResponsesWsTurnAdmission from litellm.types.google_genai.main import GenerateContentResponse from litellm.types.llms.openai_evals import ( CancelEvalResponse, @@ -7269,7 +7269,7 @@ class BaseLLMHTTPHandler: custom_llm_provider: str | None = None, first_message: str | None = None, request_defaults: ResponsesWebSocketRequestDefaults | None = None, - turn_admission: Callable[[], Awaitable[None]] | None = None, + turn_admission: "ResponsesWsTurnAdmission | None" = None, frame_guard: "ResponsesWsFrameGuard | None" = None, **kwargs: object, ) -> Exception | None: diff --git a/litellm/responses/main.py b/litellm/responses/main.py index 5c8350589b8..9259103cb5c 100644 --- a/litellm/responses/main.py +++ b/litellm/responses/main.py @@ -1,7 +1,7 @@ import asyncio import contextvars import json -from collections.abc import Awaitable, Callable, Coroutine, Generator, Iterable, Mapping, Sequence +from collections.abc import Coroutine, Generator, Iterable, Mapping, Sequence from contextlib import contextmanager from dataclasses import dataclass from functools import partial @@ -81,7 +81,7 @@ from litellm.types.llms.openai import ( # Handle ResponseText import with fallback if TYPE_CHECKING: - from litellm.responses.streaming_iterator import ResponsesWsFrameGuard + from litellm.responses.streaming_iterator import ResponsesWsFrameGuard, ResponsesWsTurnAdmission from litellm.types.llms.openai import ResponseText else: ResponseText = str # Fallback for ResponseText import @@ -2867,7 +2867,7 @@ async def _aresponses_websocket( api_base: str | None = None, api_key: str | None = None, timeout: float | None = None, - turn_admission: Callable[[], Awaitable[None]] | None = None, + turn_admission: "ResponsesWsTurnAdmission | None" = None, responses_ws_frame_guard: "ResponsesWsFrameGuard | None" = None, **kwargs, ) -> Exception | None: diff --git a/litellm/responses/streaming_iterator.py b/litellm/responses/streaming_iterator.py index f0bb2201162..fb91ba6a618 100644 --- a/litellm/responses/streaming_iterator.py +++ b/litellm/responses/streaming_iterator.py @@ -2145,18 +2145,32 @@ async def _enforce_frame_project_quota( ) -async def _admit_frame_turn(turn_admission: Callable[[], Awaitable[None]] | None, raw_message: str) -> Exception | None: +@dataclass(frozen=True, slots=True) +class _AdmittedTurn: + reservation: Mapping[str, object] = field(default_factory=lambda: MappingProxyType({})) + + +@dataclass(frozen=True, slots=True) +class _RefusedTurn: + rejection: Exception + + +ResponsesWsTurnAdmission: TypeAlias = Callable[[], Awaitable[Mapping[str, object]]] + + +async def _admit_frame_turn( + turn_admission: ResponsesWsTurnAdmission | None, raw_message: str +) -> _AdmittedTurn | _RefusedTurn: frame: Final = _json_object_or_none(raw_message) if turn_admission is None or frame is None or frame.get("type") != "response.create": - return None + return _AdmittedTurn() try: - await turn_admission() + return _AdmittedTurn(await turn_admission()) except RateLimitError as refusal: - return refusal + return _RefusedTurn(refusal) except Exception as e: verbose_logger.exception("Responses WS turn admission failed: %s", e) - return e - return None + return _RefusedTurn(e) def _turn_refusal_frame(refusal: RateLimitError) -> str: @@ -2191,6 +2205,7 @@ RESPONSES_WS_MASKABLE_TEXT_BLOCK_TYPES: Final = frozenset({"input_text", "output _RESPONSES_WS_FAILURE_EVENT_TYPES: Final = frozenset({"error", "response.failed"}) _PER_TURN_METADATA_KEYS: Final = frozenset({"metadata", "litellm_metadata"}) +_NO_IO_TOKEN_RESERVATION: Final[Mapping[str, object]] = MappingProxyType({}) _GUARDRAIL_INFORMATION_KEY: Final = "standard_logging_guardrail_information" @@ -2318,6 +2333,17 @@ def _set_turn_guardrail_information( _replace_guardrail_information(litellm_params.get(key), guardrail_information) +def _set_turn_io_token_reservation(turn_logging_obj: LiteLLMLoggingObj, reservation: Mapping[str, object]) -> None: + litellm_params: Final = _json_object_or_none( + turn_logging_obj.litellm_params # pyright: ignore[reportUnknownMemberType, reportUnknownArgumentType] # Logging.litellm_params is untyped + ) + if litellm_params is None or not reservation: + return + for bucket in (litellm_params.get(key) for key in _PER_TURN_METADATA_KEYS): + if _is_json_object(bucket): + bucket.update(reservation) + + def _replace_guardrail_information(bucket: object, guardrail_information: tuple[Mapping[str, object], ...]) -> None: if not _is_json_object(bucket): return @@ -2458,7 +2484,7 @@ class ResponsesWebSocketStreaming: authorized_model: str | None = None, custom_llm_provider: str | None = None, request_defaults: ResponsesWebSocketRequestDefaults | None = None, - turn_admission: Callable[[], Awaitable[None]] | None = None, + turn_admission: ResponsesWsTurnAdmission | None = None, frame_guard: ResponsesWsFrameGuard | None = None, ): self.websocket = websocket @@ -2480,10 +2506,13 @@ class ResponsesWebSocketStreaming: # response.create frame to prevent deployment-substitution attacks. self.authorized_model: str | None = authorized_model self.request_defaults: ResponsesWebSocketRequestDefaults | None = request_defaults - self.turn_admission: Callable[[], Awaitable[None]] | None = turn_admission + self.turn_admission: ResponsesWsTurnAdmission | None = turn_admission def _start_turn( - self, client_frame: str | None, guardrail_information: tuple[Mapping[str, object], ...] = () + self, + client_frame: str | None, + guardrail_information: tuple[Mapping[str, object], ...] = (), + io_token_reservation: Mapping[str, object] = _NO_IO_TOKEN_RESERVATION, ) -> None: start_time: Final = datetime.now() # noqa: DTZ005 # naive to match the logging pipeline start_time turn_logging_obj: Final = self.logging_obj.copy_for_new_call( @@ -2491,6 +2520,7 @@ class ResponsesWebSocketStreaming: ) self._detach_budget_reservation() _set_turn_guardrail_information(turn_logging_obj, guardrail_information) + _set_turn_io_token_reservation(turn_logging_obj, io_token_reservation) turn_logging_obj.model_call_details["messages"] = list(_logged_user_messages(client_frame)) turn_logging_obj.pre_call(input=client_frame, api_key="") self._open_turns = (*self._open_turns, _ResponsesWsTurn(logging_obj=turn_logging_obj)) @@ -2510,11 +2540,14 @@ class ResponsesWebSocketStreaming: } def _record_client_frame( - self, client_frame: str, guardrail_information: tuple[Mapping[str, object], ...] = () + self, + client_frame: str, + guardrail_information: tuple[Mapping[str, object], ...] = (), + io_token_reservation: Mapping[str, object] = _NO_IO_TOKEN_RESERVATION, ) -> None: frame: Final = _json_object_or_none(client_frame) if frame is not None and frame.get("type") == "response.create": - self._start_turn(client_frame, guardrail_information) + self._start_turn(client_frame, guardrail_information, io_token_reservation) async def _guard_frame(self, message: str) -> GuardedResponsesWsFrame | None: """Run the proxy's pre_call guardrails on one ``response.create`` frame. @@ -2533,15 +2566,18 @@ class ResponsesWebSocketStreaming: return None async def _forward_guarded_frame( - self, message: str, turn_admission: Callable[[], Awaitable[None]] | None = None + self, message: str, turn_admission: ResponsesWsTurnAdmission | None = None ) -> None: guarded: Final = await self._guard_frame(message) - if guarded is None or not await self._admit_or_reject_turn(guarded, message, turn_admission): + if guarded is None: + return + admitted: Final = await self._admit_or_reject_turn(guarded, message, turn_admission) + if admitted is None: return if guarded.pii_tokens: self._remember_pii_tokens(guarded.pii_tokens) masked: Final = await self._mask_response_create(guarded.frame) - self._record_client_frame(masked, guarded.guardrail_information) + self._record_client_frame(masked, guarded.guardrail_information, admitted.reservation) await self.backend_ws.send(masked) def _remember_pii_tokens(self, pii_tokens: Mapping[str, str]) -> None: @@ -2941,15 +2977,17 @@ class ResponsesWebSocketStreaming: return True async def _admit_or_reject_turn( - self, guarded: GuardedResponsesWsFrame, message: str, turn_admission: Callable[[], Awaitable[None]] | None - ) -> bool: - rejection: Final = await _admit_frame_turn(turn_admission, message) - if rejection is None: - return True - if self._failure is None: - self._failure = rejection - await _reject_guarded_turn(self.websocket, guarded, rejection) - return False + self, guarded: GuardedResponsesWsFrame, message: str, turn_admission: ResponsesWsTurnAdmission | None + ) -> _AdmittedTurn | None: + verdict: Final = await _admit_frame_turn(turn_admission, message) + match verdict: + case _AdmittedTurn(): + return verdict + case _RefusedTurn(rejection=rejection): + if self._failure is None: + self._failure = rejection + await _reject_guarded_turn(self.websocket, guarded, rejection) + return None async def _send_turn_refusal(self, refusal: RateLimitError) -> None: try: @@ -3052,7 +3090,7 @@ class ManagedResponsesWebSocketHandler: custom_llm_provider: str | None = None, first_message: str | None = None, quota_callbacks: Sequence[ProjectQuotaCallback] | None = None, - turn_admission: Callable[[], Awaitable[None]] | None = None, + turn_admission: ResponsesWsTurnAdmission | None = None, frame_guard: ResponsesWsFrameGuard | None = None, guardrail_callbacks: Sequence[PresidioGuardrailCallback] | None = None, **kwargs: object, @@ -3076,7 +3114,7 @@ class ManagedResponsesWebSocketHandler: self._connection_provider = self._resolve_provider(model) or custom_llm_provider self.first_message = first_message self.quota_callbacks: tuple[ProjectQuotaCallback, ...] = tuple(quota_callbacks) if quota_callbacks else () - self.turn_admission: Callable[[], Awaitable[None]] | None = turn_admission + self.turn_admission: ResponsesWsTurnAdmission | None = turn_admission self._failure: Exception | None = None # Carry through safe pass-through kwargs (e.g. extra_headers) self.extra_kwargs: dict[str, object] = {k: v for k, v in kwargs.items() if k not in _MANAGED_WS_SKIP_KWARGS} @@ -3473,7 +3511,7 @@ class ManagedResponsesWebSocketHandler: # ------------------------------------------------------------------ async def _process_response_create( - self, raw_message: str, turn_admission: Callable[[], Awaitable[None]] | None = None + self, raw_message: str, turn_admission: ResponsesWsTurnAdmission | None = None ) -> None: """ Parse one ``response.create`` event, call ``litellm.aresponses(stream=True)``, @@ -3520,11 +3558,11 @@ class ManagedResponsesWebSocketHandler: guarded: Final = await self._guard_frame(raw_message) if guarded is None: return - rejection: Final = await _admit_frame_turn(turn_admission, raw_message) - if rejection is not None: + verdict: Final = await _admit_frame_turn(turn_admission, raw_message) + if isinstance(verdict, _RefusedTurn): if self._failure is None: - self._failure = rejection - await _reject_guarded_turn(self.websocket, guarded, rejection) + self._failure = verdict.rejection + await _reject_guarded_turn(self.websocket, guarded, verdict.rejection) return if guarded.pii_tokens: self._pii_tokens = {**self._pii_tokens, **guarded.pii_tokens} @@ -3555,6 +3593,11 @@ class ManagedResponsesWebSocketHandler: ) self._update_proxy_request(call_kwargs, requested_model or self.model_group or model) call_kwargs.update(self.extra_kwargs) + turn_litellm_metadata: Final = call_kwargs.get("litellm_metadata") + call_kwargs["litellm_metadata"] = { + **(turn_litellm_metadata if _is_json_object(turn_litellm_metadata) else {}), + **verdict.reservation, + } try: terminal_event: Final = await self._stream_and_forward(model, call_kwargs) diff --git a/litellm/router.py b/litellm/router.py index af2181cedb5..233e7581a40 100644 --- a/litellm/router.py +++ b/litellm/router.py @@ -378,6 +378,7 @@ from litellm.router_utils.pre_call_checks.encrypted_content_affinity_check impor EncryptedContentAffinityCheck, ) from litellm.router_utils.pre_call_checks.io_token_rate_limit_check import ( + async_admit_io_token_turn, async_io_token_refund_failure, build_io_token_rate_limit_headers, deployment_has_io_token_limits, @@ -5113,6 +5114,7 @@ class Router(DeprecatedRouterMembers): kwargs: dict, function_name: str | None = None, splits_stream_timeout: bool = False, + opens_responses_socket: bool = False, ) -> None: """ 5 jobs: @@ -5202,7 +5204,12 @@ class Router(DeprecatedRouterMembers): # here before it's wiped below, instead of relying on that attempt's # (possibly still-pending) failure event to do it. refund_stale_reservation_before_retry(self._core_cache, kwargs) - set_io_token_rate_limit_request_kwargs(kwargs, store_in_context=deployment_has_io_token_limits(deployment)) + set_io_token_rate_limit_request_kwargs( + kwargs, + store_in_context=deployment_has_io_token_limits(deployment), + opens_responses_socket=opens_responses_socket, + metadata_channel=metadata_variable_name, + ) kwargs[metadata_variable_name].setdefault( ROUTING_REQUEST_TAGS_METADATA_KEY, @@ -6615,6 +6622,7 @@ class Router(DeprecatedRouterMembers): kwargs=kwargs, function_name=function_name, splits_stream_timeout=handler_name in _STREAM_TIMEOUT_AWARE_GENERIC_HANDLERS, + opens_responses_socket=handler_name == "_aresponses_websocket", ) data: Final = deployment["litellm_params"].copy() @@ -10465,9 +10473,20 @@ class Router(DeprecatedRouterMembers): return deployment if current is None else current.model_dump(exclude_none=True) async def async_admit_deployment_turn( - self, model_group: str, opened_on: Mapping[str, object], request_kwargs: Mapping[str, object] - ) -> None: + self, + model_group: str, + opened_on: Mapping[str, object], + request_kwargs: Mapping[str, object], + ) -> Mapping[str, object]: deployment: Final = self._deployment_as_configured_now(opened_on) + rate_limit_check: Final = next( + (callback for callback in self._routing_callbacks() if isinstance(callback, ModelRateLimitingCheck)), None + ) + io_token_admission: Final[Mapping[str, object]] = ( + MappingProxyType({}) + if rate_limit_check is None + else await async_admit_io_token_turn(rate_limit_check.dual_cache, deployment, get_utc_datetime()) + ) budget_limiter: Final = self._core_get_router_deployment_budget_limiter() if budget_limiter is not None: try: @@ -10488,6 +10507,7 @@ class Router(DeprecatedRouterMembers): await self.async_routing_strategy_pre_call_checks( # pyright: ignore[reportUnknownMemberType] # legacy signature takes an untyped deployment dict deployment=dict(without_io_token_limits(deployment)), parent_otel_span=None ) + return io_token_admission async def _admit_turn_through_routing_strategy( self, model_group: str, deployment: Mapping[str, object], request_kwargs: Mapping[str, object] diff --git a/litellm/router_utils/pre_call_checks/io_token_rate_limit_check.py b/litellm/router_utils/pre_call_checks/io_token_rate_limit_check.py index 7580c7dadd3..80e1f3fe29c 100644 --- a/litellm/router_utils/pre_call_checks/io_token_rate_limit_check.py +++ b/litellm/router_utils/pre_call_checks/io_token_rate_limit_check.py @@ -14,6 +14,9 @@ from __future__ import annotations import contextlib import contextvars from collections.abc import Mapping, MutableMapping +from dataclasses import dataclass +from datetime import datetime +from types import MappingProxyType from typing import TYPE_CHECKING, Any, Final import httpx @@ -23,7 +26,15 @@ from litellm import token_counter from litellm._internal_context import with_service_target from litellm._logging import verbose_router_logger from litellm.caching.dual_cache import DualCache +from litellm.caching.redis_cache import RedisCache, RedisCircuitBreakerOpenError from litellm.constants import NO_MAX_TOKENS_TPM_RESERVATION_FRACTION +from litellm.litellm_core_utils.internal_call_metadata import ( + IO_TOKEN_RESERVATION_KEYS, + ITPM_CACHE_KEY, + ITPM_RESERVED_KEY, + OTPM_CACHE_KEY, + OTPM_RESERVED_KEY, +) from litellm.litellm_core_utils.token_counter import deployment_openai_model, offload_token_count from litellm.llms.azure.ptu_capacity import is_str_keyed_mapping from litellm.router_utils.routing_read_batch import ROUTER_USAGE_TARGET @@ -39,18 +50,27 @@ else: RoutingArgsTTL: Final = 60 -_io_token_rate_limit_request_kwargs: Final[contextvars.ContextVar[dict[str, object] | None]] = contextvars.ContextVar( - "io_token_rate_limit_request_kwargs", + +@dataclass(frozen=True, slots=True) +class _IoTokenRateLimitRequest: + kwargs: dict[str, object] | None + opens_responses_socket: bool + metadata_channel: str + + +_io_token_rate_limit_request: Final[contextvars.ContextVar[_IoTokenRateLimitRequest | None]] = contextvars.ContextVar( + "io_token_rate_limit_request", default=None, ) -ITPM_RESERVED_KEY: Final = "_litellm_itpm_reserved" -OTPM_RESERVED_KEY: Final = "_litellm_otpm_reserved" -ITPM_CACHE_KEY: Final = "_litellm_itpm_cache_key" -OTPM_CACHE_KEY: Final = "_litellm_otpm_cache_key" - -def set_io_token_rate_limit_request_kwargs(kwargs: dict[str, object] | None, store_in_context: bool = True) -> None: +def set_io_token_rate_limit_request_kwargs( + kwargs: dict[str, object] | None, + store_in_context: bool = True, + *, + opens_responses_socket: bool = False, + metadata_channel: str, +) -> None: # The reservation sentinels are server-only, but `metadata` is caller # controlled on proxy requests. Strip any client-supplied copies here (this # runs before the router stashes its own reservation) so a forged @@ -64,11 +84,14 @@ def set_io_token_rate_limit_request_kwargs(kwargs: dict[str, object] | None, sto # every other deployment overwrite the slot with None instead of the # kwargs; overwriting (rather than skipping) also releases a previous # request's kwargs when a context is reused. - _io_token_rate_limit_request_kwargs.set(kwargs if store_in_context else None) + _io_token_rate_limit_request.set( + _IoTokenRateLimitRequest(kwargs, opens_responses_socket, metadata_channel) if store_in_context else None + ) def get_io_token_rate_limit_request_kwargs() -> dict[str, object] | None: - return _io_token_rate_limit_request_kwargs.get() + request: Final = _io_token_rate_limit_request.get() + return None if request is None else request.kwargs def seconds_until_minute_reset() -> int: @@ -224,27 +247,32 @@ def _otpm_reservation( return default if from_deployment is None else min(from_deployment, default) +def _usage_field(container: object, name: str) -> object: + if is_str_keyed_mapping(container): + return container.get(name) + return getattr(container, name, None) + + +def _cached_prompt_tokens(usage: object) -> int: + reported: Final = ( + _usage_field(_usage_field(usage, "prompt_tokens_details"), "cached_tokens"), + _usage_field(_usage_field(usage, "input_tokens_details"), "cached_tokens"), + _usage_field(usage, "cache_read_input_tokens"), + ) + return next((int(tokens) for tokens in reported if isinstance(tokens, int | float) and tokens), 0) + + def _get_usage_tokens(usage: object) -> tuple[int, int, int]: if usage is None: return 0, 0, 0 if hasattr(usage, "prompt_tokens") or hasattr(usage, "input_tokens"): prompt = int(getattr(usage, "prompt_tokens", None) or getattr(usage, "input_tokens", 0) or 0) completion = int(getattr(usage, "completion_tokens", None) or getattr(usage, "output_tokens", 0) or 0) - cached = 0 - details: object = getattr(usage, "prompt_tokens_details", None) - if details is not None: - cached = int(getattr(details, "cached_tokens", 0) or 0) - if not cached: - cached = int(getattr(usage, "cache_read_input_tokens", 0) or 0) - return prompt, completion, cached + return prompt, completion, _cached_prompt_tokens(usage) if isinstance(usage, dict): prompt = int(usage.get("prompt_tokens") or usage.get("input_tokens") or 0) completion = int(usage.get("completion_tokens") or usage.get("output_tokens") or 0) - details = usage.get("prompt_tokens_details") or {} - cached = int(details.get("cached_tokens", 0) or 0) if isinstance(details, dict) else 0 - if not cached: - cached = int(usage.get("cache_read_input_tokens") or 0) - return prompt, completion, cached + return prompt, completion, _cached_prompt_tokens(usage) return 0, 0, 0 @@ -298,19 +326,36 @@ def _resolve_reconcile_usage_tokens( if isinstance(standard_logging_object, dict): prompt_tokens = int(standard_logging_object.get("prompt_tokens") or 0) completion_tokens = int(standard_logging_object.get("completion_tokens") or 0) - cached_tokens = 0 - metadata: Final = standard_logging_object.get("metadata") - if isinstance(metadata, dict): - cached_tokens = int(metadata.get("cache_read_input_tokens") or 0) + logged_usage: Final = _usage_field(standard_logging_object.get("metadata"), "usage_object") + logged_cached_tokens: Final = _cached_prompt_tokens(logged_usage) # Same rationale as _usage_is_present: a bare total_tokens with no # prompt/completion breakdown can't be split, so it isn't treated # as resolved usage - the reservation is kept instead of refunded. if prompt_tokens or completion_tokens: - return max(0, prompt_tokens - cached_tokens), completion_tokens, True + return max(0, prompt_tokens - logged_cached_tokens), completion_tokens, True return 0, 0, False +def _reservation( + itpm_reserved: int, otpm_reserved: int, itpm_cache_key: str | None, otpm_cache_key: str | None +) -> Mapping[str, object]: + return MappingProxyType( + { + ITPM_RESERVED_KEY: itpm_reserved, + OTPM_RESERVED_KEY: otpm_reserved, + ITPM_CACHE_KEY: itpm_cache_key, + OTPM_CACHE_KEY: otpm_cache_key, + } + ) + + +def _stash_reservation_in_metadata(request: _IoTokenRateLimitRequest | None, reservation: Mapping[str, object]) -> None: + metadata: Final = None if request is None or not request.kwargs else request.kwargs.get(request.metadata_channel) + if isinstance(metadata, dict): + metadata.update(reservation) + + def _key_for_this_minute(reserved_key: str) -> str: counter, _, _ = reserved_key.rpartition(":") return f"{counter}:{get_utc_datetime().strftime('%H-%M')}" @@ -326,30 +371,6 @@ def _reconcile_increments( ) -def _stash_reservation_in_metadata( - request_kwargs: dict[str, object] | None, - *, - itpm_reserved: int, - otpm_reserved: int, - itpm_cache_key: str | None, - otpm_cache_key: str | None, -) -> None: - if not request_kwargs: - return - reservation: Final = { - ITPM_RESERVED_KEY: itpm_reserved, - OTPM_RESERVED_KEY: otpm_reserved, - ITPM_CACHE_KEY: itpm_cache_key, - OTPM_CACHE_KEY: otpm_cache_key, - } - for channel in ("metadata", "litellm_metadata"): - existing = request_kwargs.get(channel) - if isinstance(existing, dict): - existing.update(reservation) - elif channel == "metadata": - request_kwargs[channel] = dict(reservation) - - def _extract_reservation(reservation: Mapping[str, int | str | None]) -> tuple[int, int, str | None, str | None]: itpm_cache_key: Final = reservation.get(ITPM_CACHE_KEY) otpm_cache_key: Final = reservation.get(OTPM_CACHE_KEY) @@ -399,7 +420,7 @@ def _clear_reservation_from_kwargs(kwargs: Mapping[str, object] | None) -> None: for channel in _reservation_channels(kwargs): channel_dict = _as_mutable_mapping(channel) if channel_dict is not None: - for key in (ITPM_RESERVED_KEY, OTPM_RESERVED_KEY, ITPM_CACHE_KEY, OTPM_CACHE_KEY): + for key in IO_TOKEN_RESERVATION_KEYS: channel_dict.pop(key, None) @@ -499,7 +520,8 @@ def io_token_pre_call_check( if itpm_limit is None and otpm_limit is None: return deployment - request_kwargs: Final = get_io_token_rate_limit_request_kwargs() + request: Final = _io_token_rate_limit_request.get() + request_kwargs: Final = None if request is None else request.kwargs _model, served_model = _tokenizer_and_served_models(deployment) estimated_input: Final = _estimate_input_tokens(request_kwargs, model=_model) otpm_to_reserve: Final = ( @@ -546,11 +568,13 @@ def io_token_pre_call_check( raise _stash_reservation_in_metadata( - request_kwargs, - itpm_reserved=itpm_reserved, - otpm_reserved=otpm_reserved, - itpm_cache_key=itpm_key if itpm_limit is not None else None, - otpm_cache_key=otpm_key if otpm_limit is not None else None, + request, + _reservation( + itpm_reserved, + otpm_reserved, + itpm_key if itpm_limit is not None else None, + otpm_key if otpm_limit is not None else None, + ), ) return deployment @@ -565,7 +589,14 @@ async def async_io_token_pre_call_check( if itpm_limit is None and otpm_limit is None: return deployment - request_kwargs: Final = get_io_token_rate_limit_request_kwargs() + request: Final = _io_token_rate_limit_request.get() + request_kwargs: Final = None if request is None else request.kwargs + if request is not None and request.opens_responses_socket: + _stash_reservation_in_metadata( + request, + await async_admit_io_token_turn(dual_cache, deployment, get_utc_datetime(), parent_otel_span), + ) + return deployment _model, served_model = _tokenizer_and_served_models(deployment) estimated_input: Final = await offload_token_count(_estimate_input_tokens)(request_kwargs, model=_model) otpm_to_reserve: Final = ( @@ -618,15 +649,65 @@ async def async_io_token_pre_call_check( raise _stash_reservation_in_metadata( - request_kwargs, - itpm_reserved=itpm_reserved, - otpm_reserved=otpm_reserved, - itpm_cache_key=itpm_key if itpm_limit is not None else None, - otpm_cache_key=otpm_key if otpm_limit is not None else None, + request, + _reservation( + itpm_reserved, + otpm_reserved, + itpm_key if itpm_limit is not None else None, + otpm_key if otpm_limit is not None else None, + ), ) return deployment +async def _async_redis_usage(redis_cache: RedisCache | None, key: str, parent_otel_span: Span | None) -> int | None: + if redis_cache is None: + return None + with contextlib.suppress(RedisCircuitBreakerOpenError): + return await redis_cache.async_get_cache(key=key, parent_otel_span=parent_otel_span) + return None + + +@with_service_target(ROUTER_USAGE_TARGET) +async def _async_shared_usage(dual_cache: DualCache, key: str, parent_otel_span: Span | None) -> int | None: + shared: Final = await _async_redis_usage(dual_cache.redis_cache, key, parent_otel_span) + if shared is not None: + return shared + return await dual_cache.async_get_cache(key=key, local_only=True) + + +async def _raise_if_at_limit( + dual_cache: DualCache, key: str, limit: int | None, limit_label: str, parent_otel_span: Span | None +) -> None: + if limit is None: + return + usage: Final = await _async_shared_usage(dual_cache, key, parent_otel_span) + if usage is not None and usage >= limit: + raise _rate_limit_error(limit_label, limit, usage) + + +@with_service_target(ROUTER_USAGE_TARGET) +async def async_admit_io_token_turn( + dual_cache: DualCache, + deployment: Mapping[str, object], + now: datetime, + parent_otel_span: Span | None = None, +) -> Mapping[str, object]: + configured: Final = dict(deployment) + itpm_limit, otpm_limit = get_deployment_io_token_limits(configured) + if itpm_limit is None and otpm_limit is None: + return MappingProxyType({}) + cache_keys: Final = _get_cache_keys(configured, now.strftime("%H-%M")) + if cache_keys is None: + return MappingProxyType({}) + itpm_key, otpm_key = cache_keys + await _raise_if_at_limit(dual_cache, itpm_key, itpm_limit, "ITPM", parent_otel_span) + await _raise_if_at_limit(dual_cache, otpm_key, otpm_limit, "OTPM", parent_otel_span) + return _reservation( + 0, 0, itpm_key if itpm_limit is not None else None, otpm_key if otpm_limit is not None else None + ) + + @with_service_target(ROUTER_USAGE_TARGET) def io_token_reconcile_success( dual_cache: DualCache, diff --git a/tests/integration/routing/test_responses_websocket_turn_admission.py b/tests/integration/routing/test_responses_websocket_turn_admission.py index 87b56a92979..7b0e14d044f 100644 --- a/tests/integration/routing/test_responses_websocket_turn_admission.py +++ b/tests/integration/routing/test_responses_websocket_turn_admission.py @@ -518,9 +518,7 @@ def test_a_tpm_limit_lowered_while_the_socket_is_open_applies_to_its_next_turn( assert deployment.forwarded_turns() == 1 -def test_a_socket_on_an_otpm_limited_deployment_reserves_once_and_serves_every_turn( - limited: Gateway, cert: tuple[Path, Path] -) -> None: +def test_a_socket_on_an_otpm_limited_deployment_serves_every_turn(limited: Gateway, cert: tuple[Path, Path]) -> None: with responses_peer(cert) as peer, limited.scenario() as scenario: deployment: Final = _native(scenario, peer, otpm=10_000) key: Final = scenario.key(models=[deployment.name]) @@ -536,6 +534,115 @@ def test_a_socket_on_an_otpm_limited_deployment_reserves_once_and_serves_every_t assert deployment.forwarded_turns() == 5 +IO_TOKEN_LIMIT: Final = 10_000 +TOKENS_PER_TURN: Final = MappingProxyType({"itpm": USAGE["input_tokens"], "otpm": USAGE["output_tokens"]}) + + +def _io_tokens_counted(identity: str, kind: str, minute: str) -> int: + with _redis() as redis: + return sum(int(redis.get(key) or 0) for key in redis.scan_iter(f"global_router:{identity}:*:{kind}:{minute}")) + + +def _add_io_tokens(deployment: Deployment, kind: str, minute: str, tokens: int) -> None: + with _redis() as redis: + redis.incrby(f"global_router:{deployment.identity}:openai/{PROVIDER_MODEL}:{kind}:{minute}", tokens) + + +async def _say_until_counted( + connection: websockets.ClientConnection, deployment: Deployment, counted: Callable[[], float], expected: int +) -> tuple[dict[str, JsonValue], ...]: + received: Final = await _say(connection, deployment.name) + await _until(counted, lambda tokens: tokens == expected) + return received + + +@pytest.mark.parametrize("kind", ("otpm", "itpm")) +def test_a_turn_is_refused_once_earlier_turns_fill_the_deployment_io_token_limit( + limited: Gateway, cert: tuple[Path, Path], kind: str +) -> None: + per_turn: Final = TOKENS_PER_TURN[kind] + with responses_peer(cert) as peer, limited.scenario() as scenario: + deployment: Final = _native(scenario, peer, **{kind: IO_TOKEN_LIMIT}) + key: Final = scenario.key(models=[deployment.name]) + minute: Final = _fresh_minute() + counted: Final = partial(_io_tokens_counted, deployment.identity, kind, minute) + + async def session() -> tuple[tuple[dict[str, JsonValue], ...], ...]: + async with _socket(limited, key, deployment.name) as connection: + first: Final = await _say_until_counted(connection, deployment, counted, per_turn) + await asyncio.to_thread(_add_io_tokens, deployment, kind, minute, IO_TOKEN_LIMIT - 5 * per_turn) + answered: Final = tuple( + [ + await _say_until_counted( + connection, deployment, counted, IO_TOKEN_LIMIT - (5 - number) * per_turn + ) + for number in range(2, 6) + ] + ) + return (first, *answered, await _say(connection, deployment.name)) + + turns: Final = asyncio.run(session()) + + assert [completed(received)["status"] for received in turns[:5]] == ["completed"] * 5 + _assert_refused(turns[5]) + assert deployment.forwarded_turns() == 5 + assert held(counted, lambda tokens: tokens == IO_TOKEN_LIMIT, holding=3, seconds=20) == IO_TOKEN_LIMIT + + +def test_a_socket_at_half_its_deployment_otpm_serves_ten_turns(limited: Gateway, cert: tuple[Path, Path]) -> None: + with responses_peer(cert) as peer, limited.scenario() as scenario: + deployment: Final = _native(scenario, peer, otpm=IO_TOKEN_LIMIT) + key: Final = scenario.key(models=[deployment.name]) + minute: Final = _fresh_minute() + _add_io_tokens(deployment, "otpm", minute, IO_TOKEN_LIMIT // 2) + + async def session() -> tuple[tuple[dict[str, JsonValue], ...], ...]: + async with _socket(limited, key, deployment.name) as connection: + return tuple([await _say(connection, deployment.name) for _ in range(10)]) + + turns: Final = asyncio.run(session()) + + assert [completed(received)["status"] for received in turns] == ["completed"] * 10 + assert deployment.forwarded_turns() == 10 + expected: Final = IO_TOKEN_LIMIT // 2 + 10 * USAGE["output_tokens"] + counted: Final = partial(_io_tokens_counted, deployment.identity, "otpm", minute) + assert held(counted, lambda tokens: tokens == expected, holding=3, seconds=20) == expected + + +@pytest.mark.parametrize("relay", ("native", "managed")) +def test_every_turn_counts_its_actual_tokens_toward_the_deployment_itpm_and_otpm( + limited: Gateway, cert: tuple[Path, Path], relay: str +) -> None: + with responses_peer(cert) as peer, limited.scenario() as scenario: + limits: Final = {"itpm": 1_000_000, "otpm": 1_000_000} + deployment: Final = _native(scenario, peer, **limits) if relay == "native" else _managed(scenario, **limits) + key: Final = scenario.key(models=[deployment.name]) + minute: Final = _fresh_minute() + + async def session() -> tuple[tuple[dict[str, JsonValue], ...], ...]: + async with _socket(limited, key, deployment.name) as connection: + return tuple([await _say(connection, deployment.name) for _ in range(3)]) + + turns: Final = asyncio.run(session()) + + usages: Final = tuple(object_value(completed(received)["usage"]) for received in turns) + expected: Final = { + "itpm": sum(int(str(usage["input_tokens"])) for usage in usages), + "otpm": sum(int(str(usage["output_tokens"])) for usage in usages), + } + assert any(tokens > 0 for tokens in expected.values()), usages + counted: Final = { + kind: held( + partial(_io_tokens_counted, deployment.identity, kind, minute), + lambda tokens, kind=kind: tokens == expected[kind], + holding=3, + seconds=20, + ) + for kind in expected + } + assert counted == expected + + def test_usage_based_routing_refuses_a_turn_only_once_earlier_turns_used_up_the_deployment_tpm( usage_routed: Gateway, cert: tuple[Path, Path] ) -> None: diff --git a/tests/unit/responses/test_responses_websocket_all_providers.py b/tests/unit/responses/test_responses_websocket_all_providers.py index ab52e6d29c3..d0e26798090 100644 --- a/tests/unit/responses/test_responses_websocket_all_providers.py +++ b/tests/unit/responses/test_responses_websocket_all_providers.py @@ -8,15 +8,21 @@ Tests that: """ import asyncio +import itertools import json -from collections.abc import Mapping -from typing import Final -from unittest.mock import MagicMock +from collections.abc import AsyncIterator, Mapping +from datetime import datetime, timezone +from functools import partial +from typing import TYPE_CHECKING, Final +from unittest.mock import AsyncMock, MagicMock +import httpx import pytest +import respx from litellm.llms.azure.responses.transformation import AzureOpenAIResponsesAPIConfig from litellm.llms.chatgpt.responses.transformation import ChatGPTResponsesAPIConfig +from litellm.llms.custom_llm import CustomLLM from litellm.llms.databricks.responses.transformation import ( DatabricksResponsesAPIConfig, ) @@ -42,6 +48,13 @@ from litellm.llms.volcengine.responses.transformation import ( VolcEngineResponsesAPIConfig, ) from litellm.llms.xai.responses.transformation import XAIResponsesAPIConfig +from litellm.types.utils import GenericStreamingChunk + +if TYPE_CHECKING: + from litellm.caching.dual_cache import DualCache + from litellm.litellm_core_utils.litellm_logging import Logging + from litellm.responses.streaming_iterator import ResponsesWsTurnAdmission + from litellm.router_utils.pre_call_checks.model_rate_limit_check import ModelRateLimitingCheck class TestResponsesAPIWebSocketSupport: @@ -1591,7 +1604,7 @@ def _admission_refusing_the_next_turn(): from litellm.exceptions import RateLimitError - return AsyncMock(side_effect=[RateLimitError(message="crossed budget", llm_provider="", model=""), None]) + return AsyncMock(side_effect=[RateLimitError(message="crossed budget", llm_provider="", model=""), {}]) class _ClientSocket: @@ -1630,7 +1643,7 @@ class _StalledClientSocket(_ClientSocket): def _admission_crashing_the_next_turn(): from unittest.mock import AsyncMock - return AsyncMock(side_effect=[ValueError("custom check failed"), None]) + return AsyncMock(side_effect=[ValueError("custom check failed"), {}]) def _guard_blocking(word: str): @@ -1655,6 +1668,229 @@ def _guard_recording(bill_refusal: MagicMock): return guard +_OPENED_ON: Final = "global_router:ws-dep:openai/gpt-5-mini:otpm:12-34" +_TURNED_ON: Final = "global_router:ws-dep:openai/gpt-5-mini:otpm:12-35" +_OPENED_AT: Final = datetime(2026, 1, 1, 12, 34, 56, tzinfo=timezone.utc) +_TURNED_AT: Final = datetime(2026, 1, 1, 12, 35, 5, tzinfo=timezone.utc) +_WS_DEPLOYMENT: Final = { + "litellm_params": {"model": "openai/gpt-5-mini", "otpm": 4096}, + "model_info": {"id": "ws-dep"}, +} +_OTHER_USAGE: Final = 10 +_OUTPUT_TOKENS_PER_TURN: Final = 2 +_UPSTREAM_ERROR: Final = json.dumps({"type": "error", "error": {"type": "server_error", "message": "upstream failed"}}) + + +def _turn_completed() -> str: + return json.dumps( + { + "type": "response.completed", + "response": { + "id": "resp_turn", + "status": "completed", + "output": [], + "usage": {"input_tokens": 5, "output_tokens": _OUTPUT_TOKENS_PER_TURN, "total_tokens": 7}, + }, + } + ) + + +async def _socket_opened_on(dual_cache: "DualCache") -> Mapping[str, object]: + from litellm.router_utils.pre_call_checks.io_token_rate_limit_check import async_admit_io_token_turn + + await dual_cache.async_increment_cache(key=_OPENED_ON, value=_OTHER_USAGE, ttl=60) + await dual_cache.async_increment_cache(key=_TURNED_ON, value=_OTHER_USAGE, ttl=60) + return await async_admit_io_token_turn(dual_cache, _WS_DEPLOYMENT, _OPENED_AT) + + +def _turns_finish_at(monkeypatch: pytest.MonkeyPatch, finished_at: datetime) -> None: + monkeypatch.setattr( + "litellm.router_utils.pre_call_checks.io_token_rate_limit_check.get_utc_datetime", lambda: finished_at + ) + + +def _later_turns_admitted_through(dual_cache: "DualCache") -> "ResponsesWsTurnAdmission": + from litellm.router_utils.pre_call_checks.io_token_rate_limit_check import async_admit_io_token_turn + + return partial(async_admit_io_token_turn, dual_cache, _WS_DEPLOYMENT, _TURNED_AT) + + +_ITPM_OPENED_ON: Final = "global_router:ws-dep:openai/gpt-5-mini:itpm:12-34" +_ITPM_DEPLOYMENT: Final = { + "litellm_params": {"model": "openai/gpt-5-mini", "itpm": 40_000}, + "model_info": {"id": "ws-dep"}, +} +_TURN_INPUT_TOKENS: Final = 1_000 + + +def _turn_completed_reading_cache(cached_tokens: int) -> str: + return json.dumps( + { + "type": "response.completed", + "response": { + "id": "resp_cached_turn", + "status": "completed", + "output": [], + "usage": { + "input_tokens": _TURN_INPUT_TOKENS, + "input_tokens_details": {"cached_tokens": cached_tokens}, + "output_tokens": _OUTPUT_TOKENS_PER_TURN, + "total_tokens": _TURN_INPUT_TOKENS + _OUTPUT_TOKENS_PER_TURN, + }, + }, + } + ) + + +def _connection_logged_through(rate_limits: "ModelRateLimitingCheck", opened_with: Mapping[str, object]) -> "Logging": + import time + + from litellm.litellm_core_utils.litellm_logging import Logging + from litellm.types.utils import CallTypes + + connection: Final = Logging( + model="gpt-5-mini", + messages=[], + stream=False, + call_type=CallTypes.aresponses_websocket.value, + start_time=time.time(), + litellm_call_id="connection-call", + function_id="connection-call", + dynamic_async_success_callbacks=[rate_limits], + dynamic_async_failure_callbacks=[rate_limits], + ) + connection.update_environment_variables( + litellm_params={"metadata": dict(opened_with), CallTypes.aresponses_websocket.value: True}, + optional_params={}, + ) + return connection + + +def _provider_turn_stream(_request: httpx.Request) -> httpx.Response: + completed: Final = { + "id": "resp_turn", + "object": "response", + "created_at": 1767271505, + "status": "completed", + "model": "gpt-5-mini", + "output": [], + "parallel_tool_calls": True, + "tool_choice": "auto", + "tools": [], + "usage": {"input_tokens": 5, "output_tokens": _OUTPUT_TOKENS_PER_TURN, "total_tokens": 7}, + } + return httpx.Response( + 200, + headers={"content-type": "text/event-stream"}, + text="".join( + f"event: {kind}\ndata: {json.dumps({'type': kind, 'response': completed})}\n\n" + for kind in ("response.created", "response.completed") + ), + ) + + +class _ProviderResponsesSocket: + """A provider's native Responses socket that answers each ``response.create`` with a completed turn.""" + + def __init__(self) -> None: + self.sent: Final[list[dict[str, object]]] = [] + self._replies: Final[asyncio.Queue[str]] = asyncio.Queue() + + def connect(self, _url: str, **_kwargs: object) -> "_ProviderResponsesSocket": + return self + + async def __aenter__(self) -> "_ProviderResponsesSocket": + return self + + async def __aexit__(self, *_exc_info: object) -> None: + return None + + async def send(self, message: str) -> None: + self.sent.append(json.loads(message)) + await self._replies.put(_turn_completed()) + + async def recv(self, decode: bool = False) -> str: + return await self._replies.get() + + async def close(self) -> None: + return None + + +class _TwoOutputTokenTurns(CustomLLM): + def __init__(self, failing_turns: frozenset[int] = frozenset()) -> None: + super().__init__() + self._failing_turns: Final = failing_turns + self._turns: Final = itertools.count(1) + + async def astreaming(self, *args: object, **kwargs: object) -> AsyncIterator[GenericStreamingChunk]: + if next(self._turns) in self._failing_turns: + raise RuntimeError("upstream failed") + yield GenericStreamingChunk( + text="hi", + is_finished=True, + finish_reason="stop", + index=0, + tool_use=None, + usage={"prompt_tokens": 5, "completion_tokens": _OUTPUT_TOKENS_PER_TURN, "total_tokens": 7}, + ) + + +async def _run_managed_relay( + monkeypatch: pytest.MonkeyPatch, + provider: CustomLLM, + rate_limits: "ModelRateLimitingCheck", + frames: tuple[str, ...], + turn_admission: "ResponsesWsTurnAdmission | None", + opened_with: Mapping[str, object], +) -> _ClientSocket: + import litellm + + monkeypatch.setattr(litellm, "custom_provider_map", [{"provider": "ws-turn-llm", "custom_handler": provider}]) + return await _relay_turns_to(monkeypatch, "ws-turn-llm/turns", rate_limits, frames, turn_admission, opened_with) + + +async def _relay_turns_to( + monkeypatch: pytest.MonkeyPatch, + model: str, + rate_limits: "ModelRateLimitingCheck", + frames: tuple[str, ...], + turn_admission: "ResponsesWsTurnAdmission | None", + opened_with: Mapping[str, object], + **connection: object, +) -> _ClientSocket: + import litellm + from litellm.litellm_core_utils.litellm_logging import Logging + from litellm.litellm_core_utils.logging_worker import GLOBAL_LOGGING_WORKER + from litellm.responses.streaming_iterator import ManagedResponsesWebSocketHandler, drain_logging_tasks + + monkeypatch.setattr(litellm, "callbacks", [rate_limits]) + first, *later = frames + websocket: Final = _ClientSocket(*later) + handler: Final = ManagedResponsesWebSocketHandler( + websocket=websocket, + model=model, + logging_obj=Logging( + model=model, + messages=[], + stream=True, + call_type="aresponses", + start_time=0, + litellm_call_id="test-id", + function_id="test-func", + ), + litellm_metadata={"user_api_key_hash": "k1", **opened_with}, + first_message=first, + turn_admission=turn_admission, + **connection, + ) + + await handler.run() + await drain_logging_tasks() + await asyncio.sleep(0) + await asyncio.wait_for(GLOBAL_LOGGING_WORKER.flush(), timeout=10.0) + return websocket + + def _errors(websocket: _ClientSocket) -> list[tuple[object, object]]: return [(frame.get("status"), frame["error"]) for frame in websocket.sent if frame.get("type") == "error"] @@ -1722,6 +1958,195 @@ class TestWebSocketTurnAdmission: assert isinstance(failure, RateLimitError) assert "crossed budget" in str(failure) + @pytest.mark.asyncio + @pytest.mark.parametrize( + ("first_turn_end", "first_turn_tokens"), + [ + pytest.param(_turn_completed(), _OUTPUT_TOKENS_PER_TURN, id="first_turn_completed"), + pytest.param(_UPSTREAM_ERROR, 0, id="first_turn_failed"), + ], + ) + async def test_native_relay_counts_every_turn_on_the_minute_it_finishes( + self, first_turn_end: str, first_turn_tokens: int, monkeypatch: pytest.MonkeyPatch + ) -> None: + import websockets.exceptions # noqa: F401 (lazy submodule must be importable) + + from litellm.caching.dual_cache import DualCache + from litellm.responses.streaming_iterator import drain_logging_tasks + from litellm.router_utils.pre_call_checks.model_rate_limit_check import ModelRateLimitingCheck + + _turns_finish_at(monkeypatch, _TURNED_AT) + dual_cache: Final = DualCache() + opened_with: Final = await _socket_opened_on(dual_cache) + first, second, third = _turn_frames() + backend_ws: Final = MagicMock() + backend_ws.send = AsyncMock() + backend_ws.recv = AsyncMock( + side_effect=[first_turn_end, _turn_completed(), _turn_completed(), Exception("stop")] + ) + handler: Final = _make_streaming( + websocket=_ClientSocket(second, third), + backend_ws=backend_ws, + request_data={}, + logging_obj=_connection_logged_through(ModelRateLimitingCheck(dual_cache), opened_with), + first_message=first, + turn_admission=_later_turns_admitted_through(dual_cache), + ) + + await handler.client_to_backend() + await handler.backend_to_client() + await drain_logging_tasks() + + assert await dual_cache.async_get_cache(key=_OPENED_ON) == _OTHER_USAGE + assert ( + await dual_cache.async_get_cache(key=_TURNED_ON) + == _OTHER_USAGE + first_turn_tokens + 2 * _OUTPUT_TOKENS_PER_TURN + ) + + @pytest.mark.asyncio + @pytest.mark.parametrize( + ("cached_tokens", "counted_input_tokens"), + [ + pytest.param(800, _TURN_INPUT_TOKENS - 800, id="cached_input"), + pytest.param(0, _TURN_INPUT_TOKENS, id="uncached_input"), + ], + ) + async def test_native_relay_leaves_a_turn_s_cached_input_out_of_the_deployment_s_itpm( + self, cached_tokens: int, counted_input_tokens: int, monkeypatch: pytest.MonkeyPatch + ) -> None: + import websockets.exceptions # noqa: F401 (lazy submodule must be importable) + + from litellm.caching.dual_cache import DualCache + from litellm.responses.streaming_iterator import drain_logging_tasks + from litellm.router_utils.pre_call_checks.io_token_rate_limit_check import async_admit_io_token_turn + from litellm.router_utils.pre_call_checks.model_rate_limit_check import ModelRateLimitingCheck + + _turns_finish_at(monkeypatch, _OPENED_AT) + dual_cache: Final = DualCache() + opened_with: Final = await async_admit_io_token_turn(dual_cache, _ITPM_DEPLOYMENT, _OPENED_AT) + first, _, _ = _turn_frames() + backend_ws: Final = MagicMock() + backend_ws.send = AsyncMock() + backend_ws.recv = AsyncMock(side_effect=[_turn_completed_reading_cache(cached_tokens), Exception("stop")]) + handler: Final = _make_streaming( + websocket=_ClientSocket(), + backend_ws=backend_ws, + request_data={}, + logging_obj=_connection_logged_through(ModelRateLimitingCheck(dual_cache), opened_with), + first_message=first, + ) + + await handler.client_to_backend() + await handler.backend_to_client() + await drain_logging_tasks() + + assert await dual_cache.async_get_cache(key=_ITPM_OPENED_ON) == counted_input_tokens + + @pytest.mark.asyncio + @pytest.mark.parametrize( + ("failing_turns", "first_turn_tokens"), + [ + pytest.param(frozenset(), _OUTPUT_TOKENS_PER_TURN, id="first_turn_completed"), + pytest.param(frozenset({1}), 0, id="first_turn_failed"), + ], + ) + async def test_managed_relay_counts_every_turn_on_the_minute_it_finishes( + self, monkeypatch: pytest.MonkeyPatch, failing_turns: frozenset[int], first_turn_tokens: int + ) -> None: + from litellm.caching.dual_cache import DualCache + from litellm.router_utils.pre_call_checks.model_rate_limit_check import ModelRateLimitingCheck + + _turns_finish_at(monkeypatch, _TURNED_AT) + dual_cache: Final = DualCache() + opened_with: Final = await _socket_opened_on(dual_cache) + + await _run_managed_relay( + monkeypatch, + _TwoOutputTokenTurns(failing_turns), + ModelRateLimitingCheck(dual_cache), + _turn_frames(), + _later_turns_admitted_through(dual_cache), + opened_with, + ) + + assert await dual_cache.async_get_cache(key=_OPENED_ON) == _OTHER_USAGE + assert ( + await dual_cache.async_get_cache(key=_TURNED_ON) + == _OTHER_USAGE + first_turn_tokens + 2 * _OUTPUT_TOKENS_PER_TURN + ) + + @pytest.mark.asyncio + async def test_managed_relay_sends_the_provider_only_the_metadata_each_turn_carries( + self, monkeypatch: pytest.MonkeyPatch, respx_mock: respx.MockRouter + ) -> None: + import litellm + from litellm.caching.dual_cache import DualCache + from litellm.router_utils.pre_call_checks.model_rate_limit_check import ModelRateLimitingCheck + + monkeypatch.setattr(litellm, "disable_aiohttp_transport", True) + litellm.in_memory_llm_clients_cache.flush_cache() + _turns_finish_at(monkeypatch, _TURNED_AT) + provider: Final = respx_mock.post("https://api.openai.com/v1/responses").mock(side_effect=_provider_turn_stream) + client_metadata: Final = {"session": "abc", "team": "search"} + dual_cache: Final = DualCache() + + await _relay_turns_to( + monkeypatch, + "openai/gpt-5-mini", + ModelRateLimitingCheck(dual_cache), + tuple( + json.dumps({"type": "response.create", "input": word, "metadata": client_metadata}) + for word in ("one", "two") + ), + _later_turns_admitted_through(dual_cache), + await _socket_opened_on(dual_cache), + api_key="sk-fake", + ) + + assert [json.loads(call.request.read()).get("metadata") for call in provider.calls] == [ + client_metadata, + client_metadata, + ] + assert await dual_cache.async_get_cache(key=_TURNED_ON) == _OTHER_USAGE + 2 * _OUTPUT_TOKENS_PER_TURN + + @pytest.mark.asyncio + async def test_native_relay_sends_the_provider_only_the_metadata_the_socket_opened_with( + self, monkeypatch: pytest.MonkeyPatch + ) -> None: + from unittest.mock import patch + + from litellm import Router + + monkeypatch.setenv("LITELLM_RUST", "false") + provider: Final = _ProviderResponsesSocket() + client_metadata: Final = {"session": "abc", "team": "search"} + first, second, _ = _turn_frames() + router: Final = Router( + model_list=[ + { + "model_name": "ws-io", + "litellm_params": { + "model": "openai/gpt-5-mini", + "api_key": "sk-fake", + "itpm": 40_000, + "otpm": 4096, + }, + "model_info": {"id": "ws-dep"}, + } + ], + optional_pre_call_checks=["enforce_model_rate_limits"], + ) + + with patch("websockets.connect", provider.connect): + await router._aresponses_websocket( + model="ws-io", + websocket=_ClientSocket(second), + first_message=first, + metadata=dict(client_metadata), + ) + + assert [frame.get("metadata") for frame in provider.sent] == [client_metadata, client_metadata] + @pytest.mark.asyncio async def test_managed_relay_skips_a_later_turn_the_deployment_refuses(self, monkeypatch: pytest.MonkeyPatch): from unittest.mock import AsyncMock @@ -1875,7 +2300,7 @@ class TestWebSocketTurnAdmission: request_data={}, first_message=first, frame_guard=_guard_recording(bill_refusal), - turn_admission=AsyncMock(side_effect=[refusal, None]), + turn_admission=AsyncMock(side_effect=[refusal, {}]), ) close_time_failure: Final = await handler.bidirectional_forward() @@ -1904,7 +2329,7 @@ class TestWebSocketTurnAdmission: request_data={}, first_message=first, frame_guard=_guard_recording(bill_refusal), - turn_admission=AsyncMock(side_effect=[refusal, None]), + turn_admission=AsyncMock(side_effect=[refusal, {}]), ) relay: Final = asyncio.ensure_future(handler.bidirectional_forward()) await asyncio.wait_for(client.send_started.wait(), timeout=1) @@ -1955,7 +2380,7 @@ class TestWebSocketTurnAdmission: ), first_message=first, frame_guard=_guard_recording(bill_refusal), - turn_admission=AsyncMock(side_effect=[refusal, None]), + turn_admission=AsyncMock(side_effect=[refusal, {}]), ) close_time_failure: Final = await handler.run() diff --git a/tests/unit/test_router/test_io_token_rate_limits.py b/tests/unit/test_router/test_io_token_rate_limits.py index e5ff4531f84..819c6b9d4f5 100644 --- a/tests/unit/test_router/test_io_token_rate_limits.py +++ b/tests/unit/test_router/test_io_token_rate_limits.py @@ -4,14 +4,19 @@ Tests for separate ITPM/OTPM deployment rate limits (enforce_model_rate_limits). import asyncio from collections.abc import Awaitable, Callable, Mapping -from datetime import datetime +from datetime import datetime, timezone from typing import Final +from unittest.mock import AsyncMock, MagicMock import pytest +from fakeredis import FakeRedis, FakeServer +from fakeredis.aioredis import FakeRedis as FakeAsyncRedis import litellm from litellm import Router from litellm.caching.dual_cache import DualCache +from litellm.caching.in_memory_cache import InMemoryCache +from litellm.caching.redis_cache import RedisCache, RedisCircuitBreakerOpenError from litellm.constants import NO_MAX_TOKENS_TPM_RESERVATION_FRACTION from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObject from litellm.litellm_core_utils.logging_worker import GLOBAL_LOGGING_WORKER @@ -22,6 +27,7 @@ from litellm.router_utils.pre_call_checks.io_token_rate_limit_check import ( OTPM_RESERVED_KEY, _explicit_max_tokens, _reservation_value, + async_admit_io_token_turn, async_io_token_pre_call_check, async_io_token_reconcile_success, build_io_token_rate_limit_headers, @@ -35,6 +41,7 @@ from litellm.router_utils.pre_call_checks.io_token_rate_limit_check import ( from litellm.router_utils.pre_call_checks.model_rate_limit_check import ( ModelRateLimitingCheck, ) +from litellm.types.llms.openai import InputTokensDetails, ResponseAPIUsage, ResponsesAPIResponse from litellm.types.router import RouterCacheEnum from litellm.types.utils import ModelResponse, Usage from litellm.utils import get_utc_datetime @@ -115,7 +122,7 @@ class TestModelRateLimitingCheckIOTokens: "max_tokens": 10, "metadata": {}, } - set_io_token_rate_limit_request_kwargs(request_kwargs) + set_io_token_rate_limit_request_kwargs(request_kwargs, metadata_channel="metadata") await check.async_pre_call_check(deployment) minute = get_utc_datetime().strftime("%H-%M") @@ -173,7 +180,8 @@ class TestModelRateLimitingCheckIOTokens: ], "max_tokens": 10, "metadata": {}, - } + }, + metadata_channel="metadata", ) with pytest.raises(litellm.RateLimitError) as exc_info: @@ -203,7 +211,8 @@ class TestModelRateLimitingCheckIOTokens: "messages": [{"role": "user", "content": "hi"}], "max_tokens": max_tokens, "metadata": {}, - } + }, + metadata_channel="metadata", ) async def _attempt(): @@ -255,7 +264,8 @@ class TestModelRateLimitingCheckIOTokens: { "max_tokens": 5, "metadata": {}, - } + }, + metadata_channel="metadata", ) async def _attempt(): @@ -320,7 +330,7 @@ class TestModelRateLimitingCheckIOTokens: } request_kwargs = {"max_tokens": 5, "metadata": {}} - set_io_token_rate_limit_request_kwargs(request_kwargs) + set_io_token_rate_limit_request_kwargs(request_kwargs, metadata_channel="metadata") await check.async_pre_call_check(deployment) minute = get_utc_datetime().strftime("%H-%M") @@ -363,15 +373,15 @@ class TestModelRateLimitingCheckIOTokens: "model_name": "opus", } request_kwargs = {"max_tokens": 5, "metadata": {}} - set_io_token_rate_limit_request_kwargs(request_kwargs) + set_io_token_rate_limit_request_kwargs(request_kwargs, metadata_channel="metadata") await check.async_pre_call_check(deployment) # Second zero-estimate call still fits within the itpm=2 limit. - set_io_token_rate_limit_request_kwargs({"max_tokens": 5, "metadata": {}}) + set_io_token_rate_limit_request_kwargs({"max_tokens": 5, "metadata": {}}, metadata_channel="metadata") await check.async_pre_call_check(deployment) # A third exceeds the limit and is rejected. - set_io_token_rate_limit_request_kwargs({"max_tokens": 5, "metadata": {}}) + set_io_token_rate_limit_request_kwargs({"max_tokens": 5, "metadata": {}}, metadata_channel="metadata") with pytest.raises(litellm.RateLimitError): await check.async_pre_call_check(deployment) @@ -392,7 +402,7 @@ class TestModelRateLimitingCheckIOTokens: "max_tokens": 0, "metadata": {}, } - set_io_token_rate_limit_request_kwargs(zero_output_kwargs) + set_io_token_rate_limit_request_kwargs(zero_output_kwargs, metadata_channel="metadata") await check.async_pre_call_check(deployment) zero_output_otpm_key = zero_output_kwargs["metadata"][OTPM_CACHE_KEY] @@ -404,7 +414,7 @@ class TestModelRateLimitingCheckIOTokens: "max_tokens": 5, "metadata": {}, } - set_io_token_rate_limit_request_kwargs(normal_output_kwargs) + set_io_token_rate_limit_request_kwargs(normal_output_kwargs, metadata_channel="metadata") await check.async_pre_call_check(deployment) normal_output_otpm_key = normal_output_kwargs["metadata"][OTPM_CACHE_KEY] @@ -432,7 +442,7 @@ class TestModelRateLimitingCheckIOTokens: "model_name": "no-max-tokens", } request_kwargs: Final = {"messages": [{"role": "user", "content": "hello"}], "metadata": {}} - set_io_token_rate_limit_request_kwargs(request_kwargs) + set_io_token_rate_limit_request_kwargs(request_kwargs, metadata_channel="metadata") if use_sync_check: check.pre_call_check(deployment) @@ -462,7 +472,7 @@ class TestModelRateLimitingCheckIOTokens: async def _attempt() -> dict[str, object] | None: request_kwargs: Final = {"messages": [{"role": "user", "content": "hi"}], "metadata": {}} - set_io_token_rate_limit_request_kwargs(request_kwargs) + set_io_token_rate_limit_request_kwargs(request_kwargs, metadata_channel="metadata") try: await check.async_pre_call_check(deployment) except litellm.RateLimitError: @@ -493,7 +503,7 @@ class TestModelRateLimitingCheckIOTokens: "model_name": "roomy", } request_kwargs: Final = {"messages": [{"role": "user", "content": "hello"}], "metadata": {}} - set_io_token_rate_limit_request_kwargs(request_kwargs) + set_io_token_rate_limit_request_kwargs(request_kwargs, metadata_channel="metadata") await check.async_pre_call_check(deployment) @@ -507,7 +517,7 @@ class TestModelRateLimitingCheckIOTokens: async def _reserved(model_info: dict[str, str]) -> int: request_kwargs: Final = {"messages": [{"role": "user", "content": "hello"}], "metadata": {}} - set_io_token_rate_limit_request_kwargs(request_kwargs) + set_io_token_rate_limit_request_kwargs(request_kwargs, metadata_channel="metadata") deployment: Final = { "litellm_params": {"model": "azure/litellm-io-custom-deployment", "otpm": otpm_limit}, "model_info": model_info, @@ -536,7 +546,7 @@ class TestModelRateLimitingCheckIOTokens: } async def _pre_call(request_kwargs: dict[str, object]) -> None: - set_io_token_rate_limit_request_kwargs(request_kwargs) + set_io_token_rate_limit_request_kwargs(request_kwargs, metadata_channel="metadata") if use_sync_check: check.pre_call_check(deployment) else: @@ -563,12 +573,14 @@ class TestModelRateLimitingCheckIOTokens: "model_name": "opus", } request_kwargs: Final = {"messages": [{"role": "user", "content": "hello"}], "max_tokens": 100, "metadata": {}} - set_io_token_rate_limit_request_kwargs(request_kwargs) + set_io_token_rate_limit_request_kwargs(request_kwargs, metadata_channel="metadata") await check.async_pre_call_check(deployment) assert request_kwargs["metadata"][OTPM_RESERVED_KEY] == 100 - set_io_token_rate_limit_request_kwargs({"messages": [{"role": "user", "content": "hi"}], "metadata": {}}) + set_io_token_rate_limit_request_kwargs( + {"messages": [{"role": "user", "content": "hi"}], "metadata": {}}, metadata_channel="metadata" + ) with pytest.raises(litellm.RateLimitError): await check.async_pre_call_check(deployment) @@ -585,7 +597,7 @@ class TestModelRateLimitingCheckIOTokens: with_cap: Final = {"messages": [{"role": "user", "content": "hello"}], "max_tokens": 300, "metadata": {}} for request_kwargs in (without_cap, with_cap): - set_io_token_rate_limit_request_kwargs(request_kwargs) + set_io_token_rate_limit_request_kwargs(request_kwargs, metadata_channel="metadata") if use_sync_check: check.pre_call_check(deployment) else: @@ -602,7 +614,7 @@ class TestModelRateLimitingCheckIOTokens: async def reserved(deployment_cap: Mapping[str, int], request_shape: Mapping[str, object]) -> int: request_kwargs: Final = {**request_shape, "metadata": {}} - set_io_token_rate_limit_request_kwargs(request_kwargs) + set_io_token_rate_limit_request_kwargs(request_kwargs, metadata_channel="metadata") await check.async_pre_call_check( { "litellm_params": {"model": MANTLE_OPUS, "otpm": 4000, **deployment_cap}, @@ -631,7 +643,7 @@ class TestModelRateLimitingCheckIOTokens: async def reserved(deployment_cap: Mapping[str, int]) -> int: request_kwargs: Final = {"messages": [{"role": "user", "content": "hello"}], "metadata": {}} - set_io_token_rate_limit_request_kwargs(request_kwargs) + set_io_token_rate_limit_request_kwargs(request_kwargs, metadata_channel="metadata") deployment: Final = { "litellm_params": {"model": UNMAPPED_MODEL, "otpm": otpm_limit, **deployment_cap}, "model_info": {"id": f"io-deployment-cap-above-default-{deployment_cap}"}, @@ -667,7 +679,7 @@ class TestModelRateLimitingCheckIOTokens: "max_tokens": 10, "metadata": {}, } - set_io_token_rate_limit_request_kwargs(request_kwargs) + set_io_token_rate_limit_request_kwargs(request_kwargs, metadata_channel="metadata") check.pre_call_check(deployment) minute = get_utc_datetime().strftime("%H-%M") @@ -803,7 +815,7 @@ class TestModelRateLimitingCheckIOTokens: # The retry proceeds to stash deployment B's own reservation on the # same dict; it starts clean, unaffected by A's cleared sentinels. - set_io_token_rate_limit_request_kwargs(shared_kwargs) + set_io_token_rate_limit_request_kwargs(shared_kwargs, metadata_channel="metadata") itpm_key_b = f"global_router:io-retry-b:bedrock_mantle/test-b:itpm:{minute}" shared_kwargs["metadata"][ITPM_RESERVED_KEY] = 4 shared_kwargs["metadata"][ITPM_CACHE_KEY] = itpm_key_b @@ -819,7 +831,7 @@ class TestModelRateLimitingCheckIOTokens: "litellm_metadata": {OTPM_RESERVED_KEY: 7}, "litellm_params": {"metadata": {OTPM_CACHE_KEY: "attacker:key:otpm:00-00"}}, } - set_io_token_rate_limit_request_kwargs(forged) + set_io_token_rate_limit_request_kwargs(forged, metadata_channel="metadata") stored = get_io_token_rate_limit_request_kwargs() assert ITPM_RESERVED_KEY not in stored["metadata"] @@ -845,7 +857,7 @@ class TestModelRateLimitingCheckIOTokens: }, } # The router sanitizes the request kwargs before the call runs. - set_io_token_rate_limit_request_kwargs(kwargs) + set_io_token_rate_limit_request_kwargs(kwargs, metadata_channel="metadata") response = ModelResponse( choices=[{"message": {"role": "assistant", "content": "ok"}, "index": 0, "finish_reason": "stop"}], usage=Usage(prompt_tokens=1, completion_tokens=1, total_tokens=2), @@ -880,7 +892,8 @@ class TestModelRateLimitingCheckIOTokens: "messages": [{"role": "user", "content": "hello world"}], "max_tokens": 5, "metadata": {}, - } + }, + metadata_channel="metadata", ) with pytest.raises(RuntimeError): @@ -945,7 +958,7 @@ class TestModelRateLimitingCheckIOTokens: "max_tokens": 5, "metadata": {}, } - set_io_token_rate_limit_request_kwargs(request_kwargs) + set_io_token_rate_limit_request_kwargs(request_kwargs, metadata_channel="metadata") result = await async_io_token_pre_call_check(dual_cache, deployment) @@ -1147,7 +1160,7 @@ class TestModelRateLimitingCheckIOTokens: "max_tokens": 5, "metadata": {}, } - set_io_token_rate_limit_request_kwargs(request_kwargs) + set_io_token_rate_limit_request_kwargs(request_kwargs, metadata_channel="metadata") with caplog.at_level(logging.WARNING, logger="LiteLLM Router"): with pytest.raises(litellm.RateLimitError): @@ -1495,15 +1508,15 @@ class TestContextSlotRetention: "messages": [{"role": "user", "content": "x" * 1000}], "metadata": {ITPM_RESERVED_KEY: 999, ITPM_CACHE_KEY: "forged"}, } - set_io_token_rate_limit_request_kwargs(kwargs, store_in_context=False) + set_io_token_rate_limit_request_kwargs(kwargs, store_in_context=False, metadata_channel="metadata") assert get_io_token_rate_limit_request_kwargs() is None assert ITPM_RESERVED_KEY not in kwargs["metadata"] assert ITPM_CACHE_KEY not in kwargs["metadata"] - set_io_token_rate_limit_request_kwargs(kwargs, store_in_context=True) + set_io_token_rate_limit_request_kwargs(kwargs, store_in_context=True, metadata_channel="metadata") assert get_io_token_rate_limit_request_kwargs() is kwargs - set_io_token_rate_limit_request_kwargs(kwargs, store_in_context=False) + set_io_token_rate_limit_request_kwargs(kwargs, store_in_context=False, metadata_channel="metadata") assert get_io_token_rate_limit_request_kwargs() is None @pytest.mark.asyncio @@ -1516,7 +1529,7 @@ class TestContextSlotRetention: } ] ) - set_io_token_rate_limit_request_kwargs(None) + set_io_token_rate_limit_request_kwargs(None, metadata_channel="metadata") kwargs = {"messages": [{"role": "user", "content": "hello"}], "metadata": {}} deployment = router.get_deployment_by_model_group_name("plain") assert deployment is not None @@ -1534,7 +1547,7 @@ class TestContextSlotRetention: ], optional_pre_call_checks=["enforce_model_rate_limits"], ) - set_io_token_rate_limit_request_kwargs(None) + set_io_token_rate_limit_request_kwargs(None, metadata_channel="metadata") kwargs = {"messages": [{"role": "user", "content": "hello"}], "metadata": {}} deployment = router.get_deployment_by_model_group_name("limited") assert deployment is not None @@ -1551,7 +1564,7 @@ async def test_custom_azure_deployment_name_reserves_itpm_on_its_base_models_tok "max_tokens": 5, "metadata": metadata, } - set_io_token_rate_limit_request_kwargs(request_kwargs) + set_io_token_rate_limit_request_kwargs(request_kwargs, metadata_channel="metadata") await async_io_token_pre_call_check(DualCache(), deployment) return metadata[ITPM_RESERVED_KEY] @@ -1571,3 +1584,278 @@ async def test_custom_azure_deployment_name_reserves_itpm_on_its_base_models_tok ) assert named == served + + +def _reply_using(prompt_tokens: int, completion_tokens: int) -> ModelResponse: + return ModelResponse( + choices=[{"message": {"role": "assistant", "content": "hi"}, "index": 0, "finish_reason": "stop"}], + usage=Usage( + prompt_tokens=prompt_tokens, + completion_tokens=completion_tokens, + total_tokens=prompt_tokens + completion_tokens, + ), + ) + + +def _proxy_success_kwargs(reservation: Mapping[str, object]) -> dict[str, object]: + return { + "litellm_params": {"metadata": {**reservation, "user_api_key_hash": "hashed-key"}}, + "standard_logging_object": { + "hidden_params": {"litellm_model_name": "openai/gpt-5-mini"}, + "metadata": {"user_api_key_hash": "hashed-key"}, + }, + } + + +async def _deliver_success( + check: ModelRateLimitingCheck, delivery: str, kwargs: dict[str, object], reply: object +) -> None: + if delivery == "async": + await check.async_log_success_event(kwargs, reply, None, None) + return + check.log_success_event(kwargs, reply, None, None) + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "deliveries", + [("async",), ("sync",), ("async", "async"), ("async", "sync"), ("sync", "async"), ("sync", "sync")], +) +async def test_success_settles_a_reservation_held_only_in_litellm_params_metadata_exactly_once( + deliveries: tuple[str, ...], +) -> None: + dual_cache: Final = DualCache() + check: Final = ModelRateLimitingCheck(dual_cache=dual_cache) + itpm_key: Final = "global_router:proxy-shape:openai/gpt-5-mini:itpm:12-34" + otpm_key: Final = "global_router:proxy-shape:openai/gpt-5-mini:otpm:12-34" + await dual_cache.async_increment_cache(key=itpm_key, value=40, ttl=60) + await dual_cache.async_increment_cache(key=otpm_key, value=4096, ttl=60) + kwargs: Final = _proxy_success_kwargs( + {ITPM_RESERVED_KEY: 40, OTPM_RESERVED_KEY: 4096, ITPM_CACHE_KEY: itpm_key, OTPM_CACHE_KEY: otpm_key} + ) + + for delivery in deliveries: + await _deliver_success(check, delivery, kwargs, _reply_using(prompt_tokens=7, completion_tokens=3)) + + assert await dual_cache.async_get_cache(key=itpm_key) == 7 + assert await dual_cache.async_get_cache(key=otpm_key) == 3 + + +_TURN_AT: Final = datetime(2026, 1, 1, 12, 34, 56, tzinfo=timezone.utc) +_ACTUAL_TOKENS: Final = {"itpm": 5, "otpm": 2} + + +def _turn_deployment(kind: str) -> dict[str, object]: + return {"litellm_params": {"model": "openai/gpt-5-mini", kind: 100}, "model_info": {"id": f"ws-turn-{kind}"}} + + +def _turn_key(kind: str, minute: str = "12-34") -> str: + return f"global_router:ws-turn-{kind}:openai/gpt-5-mini:{kind}:{minute}" + + +@pytest.fixture +def settled_on_the_turn_s_minute(monkeypatch: pytest.MonkeyPatch) -> None: + monkeypatch.setattr( + "litellm.router_utils.pre_call_checks.io_token_rate_limit_check.get_utc_datetime", lambda: _TURN_AT + ) + + +@pytest.mark.asyncio +@pytest.mark.parametrize("kind", ["itpm", "otpm"]) +@pytest.mark.parametrize("used", [100, 101, 250]) +async def test_turn_admission_refuses_once_the_minutes_tokens_reach_the_limit(kind: str, used: int) -> None: + dual_cache: Final = DualCache() + await dual_cache.async_increment_cache(key=_turn_key(kind), value=used, ttl=60) + + with pytest.raises(litellm.RateLimitError, match=f"{kind.upper()} limit=100, current usage={used}"): + await async_admit_io_token_turn(dual_cache, _turn_deployment(kind), _TURN_AT) + + +@pytest.mark.asyncio +@pytest.mark.usefixtures("settled_on_the_turn_s_minute") +@pytest.mark.parametrize("kind", ["itpm", "otpm"]) +@pytest.mark.parametrize("used", [0, 1, 99]) +async def test_an_admitted_turn_reserves_nothing_and_counts_its_actual_tokens(kind: str, used: int) -> None: + dual_cache: Final = DualCache() + await dual_cache.async_increment_cache(key=_turn_key(kind), value=used, ttl=60) + actual: Final = _ACTUAL_TOKENS + + reservation: Final = await async_admit_io_token_turn(dual_cache, _turn_deployment(kind), _TURN_AT) + counter_key: Final = reservation[ITPM_CACHE_KEY if kind == "itpm" else OTPM_CACHE_KEY] + assert isinstance(counter_key, str) + assert await dual_cache.async_get_cache(key=counter_key) == used + await async_io_token_reconcile_success( + dual_cache, _proxy_success_kwargs(reservation), _reply_using(actual["itpm"], actual["otpm"]) + ) + + assert await dual_cache.async_get_cache(key=counter_key) == used + actual[kind] + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "deployment", + [ + {"litellm_params": {"model": "openai/gpt-5-mini", "tpm": 1}, "model_info": {"id": "ws-turn-unlimited"}}, + {"litellm_params": {"model": "openai/gpt-5-mini", "otpm": 1}}, + ], +) +async def test_a_turn_without_io_token_counters_is_admitted_with_nothing_to_settle( + deployment: dict[str, object], +) -> None: + assert await async_admit_io_token_turn(DualCache(), deployment, _TURN_AT) == {} + + +@pytest.mark.asyncio +@pytest.mark.usefixtures("settled_on_the_turn_s_minute") +@pytest.mark.parametrize("kind", ["itpm", "otpm"]) +@pytest.mark.parametrize("used_by_others", [0, 1, 99]) +async def test_a_socket_open_reserves_nothing_and_its_first_turn_counts_its_actual_tokens_on_its_own_minute( + kind: str, used_by_others: int +) -> None: + dual_cache: Final = DualCache() + await dual_cache.async_increment_cache(key=_turn_key(kind), value=used_by_others, ttl=60) + socket_open: Final[dict[str, object]] = {"metadata": {}} + set_io_token_rate_limit_request_kwargs(socket_open, opens_responses_socket=True, metadata_channel="metadata") + + assert await async_io_token_pre_call_check(dual_cache, _turn_deployment(kind)) == _turn_deployment(kind) + assert await dual_cache.async_get_cache(key=_turn_key(kind)) == used_by_others + + opened_with: Final = socket_open["metadata"] + assert isinstance(opened_with, dict) + await async_io_token_reconcile_success( + dual_cache, _proxy_success_kwargs(opened_with), _reply_using(_ACTUAL_TOKENS["itpm"], _ACTUAL_TOKENS["otpm"]) + ) + + assert await dual_cache.async_get_cache(key=_turn_key(kind)) == used_by_others + _ACTUAL_TOKENS[kind] + + +class _RedisCacheOnFakeServer(RedisCache): + def __init__(self, server: FakeServer) -> None: + super().__init__(host="io-token-turn-redis", connection_pool=FakeRedis(server=server).connection_pool) + self._fake_async_client: Final = FakeAsyncRedis(server=server) + + def init_async_client(self) -> FakeAsyncRedis: + return self._fake_async_client + + +@pytest.mark.asyncio +@pytest.mark.parametrize("kind", ["itpm", "otpm"]) +async def test_a_turn_is_refused_on_this_instance_s_count_when_reading_the_shared_counter_fails( + kind: str, +) -> None: + server: Final = FakeServer() + dual_cache: Final = DualCache(in_memory_cache=InMemoryCache(), redis_cache=_RedisCacheOnFakeServer(server)) + server.connected = False + await dual_cache.in_memory_cache.async_increment(key=_turn_key(kind), value=100) + + with pytest.raises(litellm.RateLimitError, match=f"{kind.upper()} limit=100, current usage=100"): + await async_admit_io_token_turn(dual_cache, _turn_deployment(kind), _TURN_AT) + + +@pytest.mark.asyncio +@pytest.mark.parametrize("kind", ["itpm", "otpm"]) +@pytest.mark.parametrize("delivery", ["async", "sync"]) +async def test_a_turn_finishing_in_the_next_minute_counts_its_actual_tokens_on_that_minute( + kind: str, delivery: str, monkeypatch: pytest.MonkeyPatch +) -> None: + monkeypatch.setattr( + "litellm.router_utils.pre_call_checks.io_token_rate_limit_check.get_utc_datetime", + lambda: datetime(2026, 1, 1, 12, 35, 10, tzinfo=timezone.utc), + ) + dual_cache: Final = DualCache() + reservation: Final = await async_admit_io_token_turn(dual_cache, _turn_deployment(kind), _TURN_AT) + + await _deliver_success( + ModelRateLimitingCheck(dual_cache=dual_cache), + delivery, + _proxy_success_kwargs(reservation), + _reply_using(_ACTUAL_TOKENS["itpm"], _ACTUAL_TOKENS["otpm"]), + ) + + assert [await dual_cache.async_get_cache(key=_turn_key(kind, minute)) for minute in ("12-34", "12-35")] == [ + None, + _ACTUAL_TOKENS[kind], + ] + + +@pytest.mark.asyncio +@pytest.mark.parametrize("kind", ["itpm", "otpm"]) +async def test_a_turn_is_admitted_once_the_shared_counter_drops_below_the_limit_this_instance_last_read( + kind: str, +) -> None: + redis_cache: Final = _RedisCacheOnFakeServer(FakeServer()) + dual_cache: Final = DualCache(in_memory_cache=InMemoryCache(), redis_cache=redis_cache) + await dual_cache.in_memory_cache.async_increment(key=_turn_key(kind), value=100) + await redis_cache.async_increment(key=_turn_key(kind), value=99) + + admitted: Final = await async_admit_io_token_turn(dual_cache, _turn_deployment(kind), _TURN_AT) + + assert admitted[ITPM_CACHE_KEY if kind == "itpm" else OTPM_CACHE_KEY] == _turn_key(kind) + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "responses_reply", + [ + pytest.param( + ResponsesAPIResponse( + id="resp_cached", + created_at=0, + output=[], + usage=ResponseAPIUsage( + input_tokens=50, + output_tokens=5, + total_tokens=55, + input_tokens_details=InputTokensDetails(cached_tokens=30), + ), + ), + id="usage_object", + ), + pytest.param( + { + "usage": { + "input_tokens": 50, + "output_tokens": 5, + "total_tokens": 55, + "input_tokens_details": {"cached_tokens": 30}, + } + }, + id="usage_dict", + ), + ], +) +@pytest.mark.usefixtures("settled_on_the_turn_s_minute") +async def test_a_turn_counts_only_its_uncached_responses_input_toward_itpm(responses_reply: object) -> None: + dual_cache: Final = DualCache() + await dual_cache.async_increment_cache(key=_turn_key("itpm"), value=10, ttl=60) + + reservation: Final = await async_admit_io_token_turn(dual_cache, _turn_deployment("itpm"), _TURN_AT) + await async_io_token_reconcile_success(dual_cache, _proxy_success_kwargs(reservation), responses_reply) + + assert await dual_cache.async_get_cache(key=_turn_key("itpm")) == 10 + 50 - 30 + + +def _shared_counter_answering(answer: int | Exception) -> DualCache: + redis_cache: Final = MagicMock(spec=RedisCache) + redis_cache.async_get_cache = AsyncMock(side_effect=[answer]) + return DualCache(in_memory_cache=InMemoryCache(), redis_cache=redis_cache) + + +@pytest.mark.asyncio +@pytest.mark.parametrize("kind", ["itpm", "otpm"]) +async def test_a_later_turn_is_refused_when_another_instance_filled_the_shared_minute(kind: str) -> None: + with pytest.raises(litellm.RateLimitError, match=f"{kind.upper()} limit=100, current usage=100"): + await async_admit_io_token_turn(_shared_counter_answering(100), _turn_deployment(kind), _TURN_AT) + + +@pytest.mark.asyncio +@pytest.mark.parametrize("kind", ["itpm", "otpm"]) +async def test_a_later_turn_is_admitted_on_this_instance_s_count_when_the_shared_counter_is_unreachable( + kind: str, +) -> None: + dual_cache: Final = _shared_counter_answering(RedisCircuitBreakerOpenError("circuit open")) + await dual_cache.in_memory_cache.async_increment(key=_turn_key(kind), value=99) + + admitted: Final = await async_admit_io_token_turn(dual_cache, _turn_deployment(kind), _TURN_AT) + + assert admitted[ITPM_CACHE_KEY if kind == "itpm" else OTPM_CACHE_KEY] == _turn_key(kind) diff --git a/tests/unit/test_router/test_router.py b/tests/unit/test_router/test_router.py index 1ac44b5976d..a6436e47507 100644 --- a/tests/unit/test_router/test_router.py +++ b/tests/unit/test_router/test_router.py @@ -2200,7 +2200,11 @@ async def test_ageneric_api_call_deployment_model_overrides_alias(): ) def inject_alias_into_kwargs( - deployment: dict, kwargs: dict, function_name: str | None = None, splits_stream_timeout: bool = False + deployment: dict, + kwargs: dict, + function_name: str | None = None, + splits_stream_timeout: bool = False, + opens_responses_socket: bool = False, ) -> None: # Simulate the alias leaking into kwargs (as happens when # _ageneric_api_call_with_fallbacks sets kwargs["model"] = alias before diff --git a/tests/unit/test_router/test_router_helper_utils.py b/tests/unit/test_router/test_router_helper_utils.py index 5f1b6032ca8..5f31ded626d 100644 --- a/tests/unit/test_router/test_router_helper_utils.py +++ b/tests/unit/test_router/test_router_helper_utils.py @@ -4,13 +4,15 @@ import gc import json import os import weakref -from collections.abc import Iterator, Mapping +from collections.abc import Awaitable, Callable, Iterator, Mapping from datetime import datetime, timezone from types import MethodType from typing import Final from unittest.mock import AsyncMock, MagicMock, patch +import httpx import pytest +import respx import litellm from litellm import Router @@ -538,14 +540,14 @@ async def test_deployment_callback_on_success_counts_a_responses_websocket_turn( assert await router.cache.async_get_cache(key=tpm_key) == 17 -async def _kwargs_routed_to(router: Router, model: str, function_name: str) -> dict[str, object]: +async def _kwargs_routed_to(router: Router, model: str, function_name: str, **request: object) -> dict[str, object]: routed: Final[dict[str, object]] = {} async def call(**kwargs: object) -> None: routed.update(kwargs) call.__name__ = function_name - await router._ageneric_api_call_with_fallbacks_helper(model=model, original_generic_function=call) + await router._ageneric_api_call_with_fallbacks_helper(model=model, original_generic_function=call, **request) return routed @@ -715,6 +717,363 @@ async def test_responses_websocket_turns_do_not_reserve_deployment_otpm(): await admit_turn() +def _io_limited_router(kind: str, **litellm_params: object) -> Router: + return Router( + model_list=[ + { + "model_name": "ws-io", + "litellm_params": { + "model": "openai/ws-io-unpriced", + "api_key": "sk-fake", + kind: 10_000, + **litellm_params, + }, + "model_info": {"id": "ws-io-dep"}, + } + ], + optional_pre_call_checks=["enforce_model_rate_limits"], + ) + + +def _minutes_counter_key(kind: str, minute: datetime, model: str = "openai/ws-io-unpriced") -> str: + return f"global_router:ws-io-dep:{model}:{kind}:{minute.strftime('%H-%M')}" + + +async def _count_minutes_tokens( + router: Router, kind: str, tokens: int, minute: datetime, model: str = "openai/ws-io-unpriced" +) -> None: + await router.cache.async_increment_cache(key=_minutes_counter_key(kind, minute, model), value=tokens, ttl=120) + + +async def _settle_turn( + router: Router, reservation: Mapping[str, object], input_tokens: int, output_tokens: int +) -> None: + from litellm.cost_calculator import ResponsesWebSocketTokenUsageProcessor + from litellm.litellm_core_utils.get_litellm_params import get_litellm_params + from litellm.router_utils.pre_call_checks.model_rate_limit_check import ModelRateLimitingCheck + + rate_limits: Final = next( + check for check in router.optional_callbacks or () if isinstance(check, ModelRateLimitingCheck) + ) + await rate_limits.async_log_success_event( + {"litellm_params": get_litellm_params(litellm_metadata=dict(reservation))}, + ResponsesWebSocketTokenUsageProcessor.create_logging_object( + [ + { + "type": "response.completed", + "response": {"usage": {"input_tokens": input_tokens, "output_tokens": output_tokens}}, + } + ] + ), + None, + None, + ) + + +_LONG_OUTPUT_MODEL: Final = "openai/gpt-5.4-mini" + + +def _long_output_router() -> tuple[Router, Mapping[str, int]]: + max_output: Final = litellm.get_model_info(_LONG_OUTPUT_MODEL)["max_output_tokens"] + assert isinstance(max_output, int) + limits: Final = {"itpm": 10_000, "otpm": max_output // 2} + return _io_limited_router("itpm", model=_LONG_OUTPUT_MODEL, otpm=limits["otpm"]), limits + + +@pytest.mark.asyncio +@pytest.mark.parametrize("kind", ["itpm", "otpm"]) +async def test_responses_websocket_opens_without_counting_tokens_until_the_minute_reaches_deployment_io_limit( + kind: str, router_minute_pinned: datetime +): + router, limits = _long_output_router() + + await _kwargs_routed_to(router, "ws-io", "_aresponses_websocket", websocket=object()) + + assert [ + await router.cache.async_get_cache(key=_minutes_counter_key(counted, router_minute_pinned, _LONG_OUTPUT_MODEL)) + for counted in ("itpm", "otpm") + ] == [None, None] + await _count_minutes_tokens(router, kind, limits[kind], router_minute_pinned, _LONG_OUTPUT_MODEL) + with pytest.raises(litellm.RateLimitError, match=f"{kind.upper()} limit={limits[kind]}, current usage="): + await _kwargs_routed_to(router, "ws-io", "_aresponses_websocket", websocket=object()) + + +@pytest.mark.asyncio +async def test_responses_websocket_first_turn_counts_its_actual_tokens_toward_deployment_itpm_and_otpm( + router_minute_pinned: datetime, +): + router, _ = _long_output_router() + socket_metadata: Final = (await _kwargs_routed_to(router, "ws-io", "_aresponses_websocket", websocket=object()))[ + "litellm_metadata" + ] + assert isinstance(socket_metadata, dict) + + await _settle_turn(router, socket_metadata, input_tokens=20, output_tokens=15) + + assert [ + await router.cache.async_get_cache(key=_minutes_counter_key(counted, router_minute_pinned, _LONG_OUTPUT_MODEL)) + for counted in ("itpm", "otpm") + ] == [20, 15] + + +_ADMITTED_LATE_IN_THE_MINUTE: Final = datetime(2026, 1, 1, 12, 0, 50, tzinfo=timezone.utc) +_FINISHED_IN_THE_NEXT_MINUTE: Final = datetime(2026, 1, 1, 12, 1, 15, tzinfo=timezone.utc) + + +@pytest.mark.asyncio +@pytest.mark.parametrize("kind", ["itpm", "otpm"]) +@pytest.mark.parametrize("turn", ["first", "later"]) +async def test_responses_websocket_turn_finishing_in_the_next_minute_counts_its_tokens_toward_that_minute( + kind: str, turn: str, monkeypatch: pytest.MonkeyPatch +): + _pin_router_clock(monkeypatch, _ADMITTED_LATE_IN_THE_MINUTE) + router: Final = _io_limited_router(kind) + socket: Final = await _kwargs_routed_to(router, "ws-io", "_aresponses_websocket", websocket=object()) + admit_turn: Final = socket["turn_admission"] + assert callable(admit_turn) + reservation: Final = socket["litellm_metadata"] if turn == "first" else await admit_turn() + assert isinstance(reservation, Mapping) + _pin_router_clock(monkeypatch, _FINISHED_IN_THE_NEXT_MINUTE) + + await _settle_turn(router, reservation, input_tokens=10_000, output_tokens=10_000) + + with pytest.raises(litellm.RateLimitError, match=f"{kind.upper()} limit=10000, current usage=10000"): + await admit_turn() + + +@pytest.mark.asyncio +@pytest.mark.parametrize("body", [{}, {"websocket": True}], ids=["plain", "websocket_key"]) +async def test_messages_call_reserves_max_tokens_against_deployment_otpm_whatever_websocket_key_its_body_carries( + body: Mapping[str, object], router_minute_pinned: datetime +): + router: Final = _io_limited_router("otpm") + + for _ in range(3): + await _kwargs_routed_to(router, "ws-io", "anthropic_messages", max_tokens=3_000, **body) + + assert await router.cache.async_get_cache(key=_minutes_counter_key("otpm", router_minute_pinned)) == 9_000 + with pytest.raises(litellm.RateLimitError, match="OTPM limit=10000, current usage=12000"): + await _kwargs_routed_to(router, "ws-io", "anthropic_messages", max_tokens=3_000, **body) + + +async def _realtime_opens_until_refused(monkeypatch: pytest.MonkeyPatch, **socket: object) -> tuple[int, str]: + monkeypatch.setattr(litellm, "callbacks", []) + router: Final = _io_limited_router("otpm", otpm=50) + for opened in range(51): + try: + await _kwargs_routed_to(router, "ws-io", "_arealtime", **socket) + except litellm.RateLimitError as refused: + return opened, str(refused) + pytest.fail("realtime opens never reached the deployment's otpm") + + +@pytest.mark.asyncio +@pytest.mark.usefixtures("router_minute_pinned") +async def test_realtime_open_reserves_and_is_refused_at_deployment_otpm_like_an_open_without_a_socket( + monkeypatch: pytest.MonkeyPatch, +): + with_socket: Final = await _realtime_opens_until_refused(monkeypatch, websocket=object()) + + assert "OTPM limit=50, current usage=" in with_socket[1] + assert with_socket == await _realtime_opens_until_refused(monkeypatch) + + +@pytest.mark.asyncio +@pytest.mark.parametrize("kind", ["itpm", "otpm"]) +async def test_responses_websocket_turn_is_refused_once_the_minutes_tokens_reach_deployment_io_limit( + kind: str, router_minute_pinned: datetime +): + router: Final = _io_limited_router(kind) + admit_turn: Final = (await _kwargs_routed_to(router, "ws-io", "_aresponses_websocket"))["turn_admission"] + assert callable(admit_turn) + await _count_minutes_tokens(router, kind, 10_000, router_minute_pinned) + + with pytest.raises(litellm.RateLimitError, match=f"{kind.upper()} limit=10000, current usage="): + await admit_turn() + + +@pytest.mark.asyncio +@pytest.mark.parametrize("kind", ["itpm", "otpm"]) +async def test_responses_websocket_turn_refused_for_tokens_leaves_the_daily_request_limit_unused( + kind: str, router_minute_pinned: datetime +): + router: Final = _io_limited_router(kind, rpd=2) + admit_turn: Final = (await _kwargs_routed_to(router, "ws-io", "_aresponses_websocket"))["turn_admission"] + assert callable(admit_turn) + await _count_minutes_tokens(router, kind, 10_000, router_minute_pinned) + with pytest.raises(litellm.RateLimitError, match=f"{kind.upper()} limit=10000"): + await admit_turn() + await _count_minutes_tokens(router, kind, -10_000, router_minute_pinned) + + await admit_turn() + + with pytest.raises(litellm.RateLimitError, match="RPD"): + await admit_turn() + + +_PROVIDER_RESPONSE: Final = { + "id": "resp_io", + "object": "response", + "created_at": 1767268830, + "status": "completed", + "model": "ws-io-unpriced", + "output": [ + { + "type": "message", + "id": "msg_io", + "status": "completed", + "role": "assistant", + "content": [{"type": "output_text", "text": "Done.", "annotations": []}], + } + ], + "parallel_tool_calls": True, + "tool_choice": "auto", + "tools": [], + "usage": {"input_tokens": 10, "output_tokens": 5, "total_tokens": 15}, +} + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "client_metadata", [None, {"session": "abc", "team": "search"}], ids=["without_metadata", "with_metadata"] +) +async def test_responses_call_sends_the_provider_only_the_client_metadata_and_settles_deployment_io_tokens( + client_metadata: Mapping[str, str] | None, + router_minute_pinned: datetime, + monkeypatch: pytest.MonkeyPatch, + respx_mock: respx.MockRouter, +): + from litellm.litellm_core_utils.logging_worker import GLOBAL_LOGGING_WORKER + from litellm.responses.streaming_iterator import drain_logging_tasks + + monkeypatch.setattr(litellm, "disable_aiohttp_transport", True) + litellm.in_memory_llm_clients_cache.flush_cache() + provider: Final = respx_mock.post("https://api.openai.com/v1/responses").mock( + return_value=httpx.Response(200, json=_PROVIDER_RESPONSE) + ) + router: Final = _io_limited_router("itpm", otpm=10_000) + + await router.aresponses( + model="ws-io", + input="hi", + max_output_tokens=50, + **({} if client_metadata is None else {"metadata": dict(client_metadata)}), + ) + await drain_logging_tasks() + await asyncio.wait_for(GLOBAL_LOGGING_WORKER.flush(), timeout=10.0) + + assert [json.loads(call.request.read()).get("metadata") for call in provider.calls] == [client_metadata] + assert [ + await router.cache.async_get_cache(key=_minutes_counter_key(kind, router_minute_pinned)) + for kind in ("itpm", "otpm") + ] == [10, 5] + + +@pytest.mark.parametrize( + "client_metadata", [None, {"session": "abc", "team": "search"}], ids=["without_metadata", "with_metadata"] +) +def test_sync_responses_call_sends_the_provider_only_the_client_metadata( + client_metadata: Mapping[str, str] | None, monkeypatch: pytest.MonkeyPatch, respx_mock: respx.MockRouter +): + litellm.in_memory_llm_clients_cache.flush_cache() + provider: Final = respx_mock.post("https://api.openai.com/v1/responses").mock( + return_value=httpx.Response(200, json=_PROVIDER_RESPONSE) + ) + + _io_limited_router("itpm", otpm=10_000).responses( + model="ws-io", + input="hi", + max_output_tokens=50, + **({} if client_metadata is None else {"metadata": dict(client_metadata)}), + ) + + assert [json.loads(call.request.read()).get("metadata") for call in provider.calls] == [client_metadata] + + +_ANTHROPIC_PROVIDER_RESPONSE: Final = { + "id": "msg_io", + "type": "message", + "role": "assistant", + "model": "ws-io-unpriced", + "content": [{"type": "text", "text": "Done."}], + "stop_reason": "end_turn", + "stop_sequence": None, + "usage": {"input_tokens": 10, "output_tokens": 5}, +} + + +async def _responses_call(router: Router) -> object: + return await router.aresponses(model="ws-io", input="hi", max_output_tokens=50) + + +async def _messages_call(router: Router) -> object: + return await router.aanthropic_messages(model="ws-io", messages=[{"role": "user", "content": "hi"}], max_tokens=50) + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + ("model", "provider_url", "provider_response", "call"), + [ + pytest.param( + "openai/ws-io-unpriced", + "https://api.openai.com/v1/responses", + _PROVIDER_RESPONSE, + _responses_call, + id="responses", + ), + pytest.param( + "anthropic/ws-io-unpriced", + "https://api.anthropic.com/v1/messages", + _ANTHROPIC_PROVIDER_RESPONSE, + _messages_call, + id="messages", + ), + ], +) +async def test_retried_call_settles_deployment_io_tokens_to_the_successful_attempt_s_tokens( + model: str, + provider_url: str, + provider_response: Mapping[str, object], + call: Callable[[Router], Awaitable[object]], + router_minute_pinned: datetime, + monkeypatch: pytest.MonkeyPatch, + respx_mock: respx.MockRouter, +): + from litellm.litellm_core_utils.logging_worker import GLOBAL_LOGGING_WORKER + from litellm.responses.streaming_iterator import drain_logging_tasks + + monkeypatch.setattr(litellm, "disable_aiohttp_transport", True) + litellm.in_memory_llm_clients_cache.flush_cache() + provider: Final = respx_mock.post(provider_url).mock( + side_effect=[ + httpx.Response(500, json={"error": {"message": "upstream failed", "type": "server_error"}}), + httpx.Response(200, json=provider_response), + ] + ) + router: Final = Router( + model_list=[ + { + "model_name": "ws-io", + "litellm_params": {"model": model, "api_key": "sk-fake", "itpm": 10_000, "otpm": 10_000}, + "model_info": {"id": "ws-io-dep"}, + } + ], + optional_pre_call_checks=["enforce_model_rate_limits"], + num_retries=1, + retry_after=0, + ) + + await call(router) + await drain_logging_tasks() + await asyncio.wait_for(GLOBAL_LOGGING_WORKER.flush(), timeout=10.0) + + assert provider.call_count == 2 + assert [ + await router.cache.async_get_cache(key=_minutes_counter_key(kind, router_minute_pinned, model)) + for kind in ("itpm", "otpm") + ] == [10, 5] + + @pytest.mark.asyncio async def test_deployment_callback_on_success_tracks_tpm_for_io_deployment(): """ @@ -1194,10 +1553,17 @@ def _rpm_tpm_router(model_id: str) -> Router: ) +def _pin_router_clock(monkeypatch: pytest.MonkeyPatch, pinned: datetime) -> None: + monkeypatch.setattr("litellm.router.get_utc_datetime", lambda: pinned) + monkeypatch.setattr( + "litellm.router_utils.pre_call_checks.io_token_rate_limit_check.get_utc_datetime", lambda: pinned + ) + + @pytest.fixture def router_minute_pinned(monkeypatch: pytest.MonkeyPatch) -> datetime: pinned: Final = datetime(2026, 1, 1, 12, 0, 30, tzinfo=timezone.utc) - monkeypatch.setattr("litellm.router.get_utc_datetime", lambda: pinned) + _pin_router_clock(monkeypatch, pinned) return pinned From df47ec01717cef4e5614d34081a568be4e5de4c6 Mon Sep 17 00:00:00 2001 From: quietjules <349048734+quietjules@users.noreply.github.com> Date: Fri, 30 Oct 2026 08:27:35 -0700 Subject: [PATCH 06/18] fix(spend): keep a NaN or infinite cost from wedging daily spend writes (#48829) * fix(spend): keep a NaN or infinite cost from wedging daily spend writes A non-finite provider-reported cost now counts as not reported, so the token price bills the request. Any other non-finite cost is logged and recorded as 0, and spend log JSON writes NaN and Infinity as null so Postgres accepts the row. * fix(spend): price non-finite provider costs on every route and drain parked NaN batches Perplexity, OpenRouter pass-through, and Responses streaming now treat a NaN or infinite reported cost as unreported, so the token price bills the request. A daily spend batch parked in the Redis buffer by an older version with a NaN spend now drains as 0 instead of failing every flush. The cost header never carries NaN or Infinity, and a set holding a non-finite float no longer breaks spend log JSON * fix(spend): record non-finite savings as 0 and warn when a parked NaN batch drains An overflowing price on a compressed request made compression savings infinite, which still wedged the daily spend batch. Every savings column now records a non-finite value as 0 with a warning. The Redis drain warns when it reads a NaN an older version parked, since the spend summed into it is lost from daily spend. The writer test now runs the real payload builder and spend log insert * fix(spend): keep non-finite costs out of headers, responses, and admin JSON Writing NaN as null is now opt-in for spend log columns, the Redis buffer, and streamed frames, so admin writes and logging integrations keep their output. The cost calculator records an overflowing total as 0 with a warning, cost headers leave out non-finite values, a non-streaming NaN usage.cost is dropped instead of failing the response, and a parked NaN spend log row reads back as 0. * fix(spend): keep caller-set call ids from splitting the non-finite cost warning The overflow warning strips newlines from litellm_call_id, and neither non-finite cost warning prints the raw value, which said nothing beyond "non-finite" * fix(spend): drop non-finite Responses and OpenRouter costs and the breakdown of an unpriceable request * fix(types): drop a usage cost breakdown that holds a non-finite number * fix(streaming): keep a dropped non-finite final usage cost from falling back to an earlier chunk's cost * fix(spend): keep the dropped-cost flag out of reach of provider usage keys A provider usage key named dropped_non_finite_cost shadowed the Usage method on the instance, so the stream builder raised and dropped the request's spend. The builder now reads the flag through the class, and Usage sets the flag after copying provider keys, so a _dropped_non_finite_cost key cannot set it either. --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> --- litellm/litellm_core_utils/core_helpers.py | 21 ++- litellm/litellm_core_utils/litellm_logging.py | 25 ++- litellm/litellm_core_utils/safe_json_dumps.py | 23 ++- .../streaming_chunk_builder_utils.py | 9 +- .../llms/openrouter/chat/transformation.py | 8 +- .../pass_through/typesafe_logging_handler.py | 6 +- litellm/llms/perplexity/cost_calculator.py | 8 +- litellm/proxy/common_request_processing.py | 15 +- litellm/proxy/db/db_spend_update_writer.py | 12 +- .../redis_update_buffer.py | 24 ++- .../proxy/hooks/proxy_track_cost_callback.py | 3 +- litellm/proxy/spend_tracking/savings.py | 6 +- .../spend_tracking/spend_tracking_utils.py | 11 +- litellm/types/llms/openai.py | 10 +- litellm/types/utils.py | 30 +++- .../test_non_finite_cost_spend_writes.py | 166 ++++++++++++++++++ .../test_convert_dict_to_response.py | 44 +++++ .../litellm_core_utils/test_core_helpers.py | 28 ++- .../test_litellm_logging.py | 54 ++++++ .../test_safe_json_dumps.py | 25 +++ .../test_streaming_chunk_builder_utils.py | 41 +++++ .../test_openrouter_chat_transformation.py | 29 +++ ...est_openrouter_responses_transformation.py | 25 +++ .../test_perplexity_cost_calculator.py | 22 +++ .../test_redis_update_buffer.py | 130 ++++++++++++++ .../proxy/db/test_db_spend_update_writer.py | 54 ++++++ ...st_typesafe_passthrough_logging_handler.py | 29 +++ .../unit/proxy/spend_tracking/test_savings.py | 28 +++ .../proxy/test_common_request_processing.py | 49 ++++++ tests/unit/test_main.py | 22 +++ 30 files changed, 906 insertions(+), 51 deletions(-) create mode 100644 tests/integration/spend/test_non_finite_cost_spend_writes.py diff --git a/litellm/litellm_core_utils/core_helpers.py b/litellm/litellm_core_utils/core_helpers.py index 28f45d2bb2c..2f81154ab83 100644 --- a/litellm/litellm_core_utils/core_helpers.py +++ b/litellm/litellm_core_utils/core_helpers.py @@ -982,25 +982,36 @@ def _usage_field(usage: object, name: str) -> object: return getattr(usage, name, None) -def _cost_number(value: object) -> float | None: - if isinstance(value, bool) or not isinstance(value, int | float): +def finite_cost(value: object) -> float | None: + if isinstance(value, bool) or not isinstance(value, int | float) or not math.isfinite(value): return None return float(value) +def single_log_line(value: object) -> str: + return str(value).replace("\r", "").replace("\n", "") + + +def finite_reported_cost(value: object) -> float | None: + cost: Final = finite_cost(value) + if cost is None and isinstance(value, float): + verbose_logger.warning("Ignoring a non-finite provider-reported cost %s", value) + return cost + + def _reported_cost_value(value: object) -> float | None: """A reported cost is a number, or a breakdown whose total lives under ``total_cost``.""" raw: Final[object] = value if isinstance(value, Mapping): - return _cost_number(_usage_field(raw, "total_cost")) - return _cost_number(value) + return finite_reported_cost(_usage_field(raw, "total_cost")) + return finite_reported_cost(value) def response_cost_in_hidden_params(response: _CarriesHiddenParams) -> float | None: additional_headers: Final[object] = response.hidden_params.get("additional_headers") if not isinstance(additional_headers, Mapping): return None - return _cost_number(_USAGE_BLOCK_ADAPTER.validate_python(additional_headers).get(RESPONSE_COST_HEADER)) + return finite_reported_cost(_USAGE_BLOCK_ADAPTER.validate_python(additional_headers).get(RESPONSE_COST_HEADER)) def override_stamped_response_cost(response: _CarriesHiddenParams, cost: float) -> None: diff --git a/litellm/litellm_core_utils/litellm_logging.py b/litellm/litellm_core_utils/litellm_logging.py index 8bfb0b60693..cfe8c1739c6 100644 --- a/litellm/litellm_core_utils/litellm_logging.py +++ b/litellm/litellm_core_utils/litellm_logging.py @@ -76,6 +76,7 @@ from litellm.litellm_core_utils.classifier_logging import ( is_classifier_call, ) from litellm.litellm_core_utils.core_helpers import ( + finite_cost, get_provider_response_headers_from_hidden_params, is_expected_client_error, native_messages_reported_cost, @@ -84,6 +85,7 @@ from litellm.litellm_core_utils.core_helpers import ( proxy_stamped_used_client_oauth_token, reconstruct_model_name, set_response_cost_in_hidden_params, + single_log_line, ) from litellm.litellm_core_utils.error_normalization import normalize_error from litellm.litellm_core_utils.get_litellm_params import get_litellm_params @@ -2093,7 +2095,7 @@ class Logging(LiteLLMLoggingBaseClass): litellm_model_name=litellm_model_name, router_model_id=router_model_id or hidden_params.get("model_id"), ) - return hidden_params["response_cost"] + return self._recorded_response_cost(hidden_params["response_cost"]) elif router_model_id is None and "model_id" in hidden_params: # use model_id if not already set router_model_id = hidden_params["model_id"] @@ -2193,7 +2195,7 @@ class Logging(LiteLLMLoggingBaseClass): litellm_model_name=litellm_model_name, router_model_id=router_model_id, ) - return total_response_cost + return self._recorded_response_cost(total_response_cost) except Exception as e: # error calculating cost debug_info = StandardLoggingModelCostFailureDebugInformation( error_str=str(e), @@ -2219,6 +2221,15 @@ class Logging(LiteLLMLoggingBaseClass): _response_cost_calculator = response_cost_calculator + def _recorded_response_cost(self, cost: float | None) -> float | None: + if cost is None or finite_cost(cost) is not None: + return cost + verbose_logger.warning( + "Recording a non-finite response_cost as 0 for litellm_call_id=%s", single_log_line(self.litellm_call_id) + ) + self.cost_breakdown = None + return 0.0 + def uncached_price(self, result: object) -> UncachedPrice: previous_cost_breakdown: Final = self.cost_breakdown previous_billed_token_rates: Final = self.billed_token_rates @@ -7582,7 +7593,15 @@ def get_standard_logging_object_payload( ) custom_pricing: Final = False if spilled_over else logging_obj.custom_pricing_for(init_response_obj) raw_response_cost: Final = kwargs.get("response_cost") - llm_response_cost: Final[float] = raw_response_cost or 0.0 + numeric_response_cost: Final = raw_response_cost if isinstance(raw_response_cost, int | float) else None + priced_response_cost: Final = finite_cost(numeric_response_cost) + if numeric_response_cost is not None and priced_response_cost is None: + verbose_logger.warning( + "Logging a non-finite response_cost=%s as 0 for litellm_call_id=%s", + numeric_response_cost, + single_log_line(logging_obj.litellm_call_id), + ) + llm_response_cost: Final[float] = priced_response_cost or 0.0 guardrail_cost: Final = guardrail_information_cost(metadata.get("standard_logging_guardrail_information")) finalized_cost: Final = _finalized_cost_of(logging_obj) response_cost: Final[float] = ( diff --git a/litellm/litellm_core_utils/safe_json_dumps.py b/litellm/litellm_core_utils/safe_json_dumps.py index 24c401281ef..909e1222ae6 100644 --- a/litellm/litellm_core_utils/safe_json_dumps.py +++ b/litellm/litellm_core_utils/safe_json_dumps.py @@ -1,4 +1,5 @@ import json +import math from collections.abc import Callable from typing import Any, Final @@ -14,16 +15,23 @@ def strip_null_bytes(value: str) -> str: return value.replace("\x00", "") +def _finite_or_none(value: float | None) -> float | None: + return None if isinstance(value, float) and not math.isfinite(value) else value + + def safe_json_structure( data: object, max_depth: int = DEFAULT_MAX_RECURSE_DEPTH, value_transform: Callable[[str | None, str], str] | None = None, key: str | None = None, + *, + non_finite_as_null: bool = False, ) -> object: """ Rebuild data out of JSON-native pieces while detecting circular references. If a circular reference is detected then a marker string is returned. - NUL bytes are stripped from strings to prevent PostgreSQL 22P05 errors. + NUL bytes are stripped from strings to prevent PostgreSQL 22P05 errors. With + non_finite_as_null, NaN and infinite floats become null, since jsonb rejects them. value_transform, when given, is applied to every string leaf (and to the str() fallback for non-serializable objects) with the mapping key the leaf @@ -43,7 +51,7 @@ def safe_json_structure( cleaned = obj.replace("\x00", "") if "\x00" in obj else obj return _transform(key, cleaned) if isinstance(obj, (int, float, bool, type(None))): - return obj + return _finite_or_none(obj) if non_finite_as_null else obj # Check for circular reference. if id(obj) in seen: return "CircularReference Detected" @@ -65,8 +73,9 @@ def safe_json_structure( seen.remove(id(obj)) return tuple_result elif isinstance(obj, set): - sortable_values: Final[list[Any]] = [_serialize(item, seen, depth + 1, key) for item in obj] - set_result: Final = sorted(sortable_values) + members: Final[list[Any]] = [_serialize(item, seen, depth + 1, key) for item in obj] + present: Final = sorted(member for member in members if member is not None) + set_result: Final = present + [None] * members.count(None) seen.remove(id(obj)) return set_result elif isinstance(obj, BaseModel): @@ -88,6 +97,10 @@ def safe_dumps( data: object, max_depth: int = DEFAULT_MAX_RECURSE_DEPTH, value_transform: Callable[[str | None, str], str] | None = None, + *, + non_finite_as_null: bool = False, ) -> str: """Serialize data to JSON text through safe_json_structure.""" - return json.dumps(safe_json_structure(data, max_depth, value_transform), default=str) + return json.dumps( + safe_json_structure(data, max_depth, value_transform, non_finite_as_null=non_finite_as_null), default=str + ) diff --git a/litellm/litellm_core_utils/streaming_chunk_builder_utils.py b/litellm/litellm_core_utils/streaming_chunk_builder_utils.py index 94351e72ab1..574f1103c1a 100644 --- a/litellm/litellm_core_utils/streaming_chunk_builder_utils.py +++ b/litellm/litellm_core_utils/streaming_chunk_builder_utils.py @@ -15,6 +15,7 @@ from litellm._logging import verbose_logger from litellm.litellm_core_utils.core_helpers import ( PROVIDERS_BILLED_FROM_THEIR_OWN_USAGE_FIELDS, chunk_usage, + finite_cost, provider_reported_usage_cost, stamp_litellm_bill_on_usage, ) @@ -1014,7 +1015,9 @@ class ChunkProcessor: prompt_tokens_details, cache_creation_token_details ) - if usage_chunk_dict["cost"] is not None: + if Usage.dropped_non_finite_cost(usage_chunk): + cost = None + elif usage_chunk_dict["cost"] is not None: cost = usage_chunk_dict["cost"] prompt_tokens_details = attach_cache_creation_token_details(prompt_tokens_details, cache_creation_token_details) @@ -1444,8 +1447,8 @@ def _set_stream_builder_response_cost( def _stamp_streaming_usage_cost(usage: Usage, response: ModelResponse, logging_obj: Optional["Logging"]) -> None: if logging_obj is None: return - if _logged_provider(logging_obj) in PROVIDERS_BILLED_FROM_THEIR_OWN_USAGE_FIELDS and isinstance( - getattr(usage, "cost", None), (int, float) + if _logged_provider(logging_obj) in PROVIDERS_BILLED_FROM_THEIR_OWN_USAGE_FIELDS and ( + finite_cost(getattr(usage, "cost", None)) is not None ): return computed_cost: Final = logging_obj.response_cost_calculator(result=response) diff --git a/litellm/llms/openrouter/chat/transformation.py b/litellm/llms/openrouter/chat/transformation.py index 9aaf552e79a..cca6815399f 100644 --- a/litellm/llms/openrouter/chat/transformation.py +++ b/litellm/llms/openrouter/chat/transformation.py @@ -13,6 +13,7 @@ from typing import TYPE_CHECKING, Final, cast import httpx from pydantic import ConfigDict, TypeAdapter +from litellm.litellm_core_utils.core_helpers import finite_cost from litellm.litellm_core_utils.hidden_params import HIDDEN_PARAMS_ATTR, set_hidden_params from litellm.llms.base_llm.base_model_iterator import BaseModelResponseIterator from litellm.llms.base_llm.chat.transformation import BaseLLMException @@ -225,8 +226,11 @@ class OpenrouterConfig(OpenAIGPTConfig): additional_headers: Final = cast( # cast-ok: preserve mapping operations on response metadata dict[str, object], hidden_params["additional_headers"] ) - if isinstance(response_cost, str | int | float): - additional_headers["llm_provider-x-litellm-response-cost"] = float(response_cost) + reported_cost: Final = ( + finite_cost(float(response_cost)) if isinstance(response_cost, str | int | float) else None + ) + if reported_cost is not None: + additional_headers["llm_provider-x-litellm-response-cost"] = reported_cost except Exception: # If we can't extract cost, continue without it - don't fail the response pass diff --git a/litellm/llms/pass_through/typesafe_logging_handler.py b/litellm/llms/pass_through/typesafe_logging_handler.py index 9eae57774c2..7a41d2a5d14 100644 --- a/litellm/llms/pass_through/typesafe_logging_handler.py +++ b/litellm/llms/pass_through/typesafe_logging_handler.py @@ -6,6 +6,7 @@ import httpx from pydantic import TypeAdapter, ValidationError import litellm +from litellm.litellm_core_utils.core_helpers import finite_reported_cost from litellm.litellm_core_utils.decisions_pricing import apply_decisions_pricing from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj from litellm.litellm_core_utils.litellm_logging import ( @@ -100,9 +101,10 @@ class TypeSafePassthroughLoggingHandler: f"{custom_llm_provider}/{model}" for model in candidate_models if model is not None ) pricing: Final = _pricing_for(candidate_model_keys) + reported_cost: Final = finite_reported_cost(usage.cost) if custom_llm_provider == "openrouter" else None response_cost: Final = ( - usage.cost - if custom_llm_provider == "openrouter" and usage.cost is not None + reported_cost + if reported_cost is not None else input_tokens * pricing.input_cost_per_token + output_tokens * pricing.output_cost_per_token ) usage_object: Final = Usage( diff --git a/litellm/llms/perplexity/cost_calculator.py b/litellm/llms/perplexity/cost_calculator.py index 2d762adf958..c3d178771a5 100644 --- a/litellm/llms/perplexity/cost_calculator.py +++ b/litellm/llms/perplexity/cost_calculator.py @@ -6,6 +6,7 @@ Helper util for handling perplexity-specific cost calculation from datetime import datetime from typing import Final +from litellm.litellm_core_utils.core_helpers import finite_reported_cost from litellm.litellm_core_utils.llm_cost_calc.utils import TokenRates, apply_off_peak_pricing from litellm.types.utils import Usage from litellm.utils import get_model_info @@ -34,10 +35,9 @@ def cost_per_token(model: str, usage: Usage, current_time: datetime | None = Non total_cost = cost_info.get("total_cost") elif isinstance(cost_info, (int, float)) and not isinstance(cost_info, bool): total_cost = float(cost_info) - if total_cost is not None: - # Return total cost as completion_cost (prompt_cost=0) since Perplexity - # doesn't break down by input/output in their cost object - return (0.0, float(total_cost)) + reported_cost: Final = None if total_cost is None else finite_reported_cost(float(total_cost)) + if reported_cost is not None: + return (0.0, reported_cost) ## FALLBACK: Calculate cost manually if Perplexity doesn't provide it ## GET MODEL INFO diff --git a/litellm/proxy/common_request_processing.py b/litellm/proxy/common_request_processing.py index 48eb58adaf2..762f552dadc 100644 --- a/litellm/proxy/common_request_processing.py +++ b/litellm/proxy/common_request_processing.py @@ -60,6 +60,7 @@ from litellm.litellm_core_utils.bug_report import ( strip_bug_report_notice, ) from litellm.litellm_core_utils.core_helpers import ( + finite_cost, get_or_create_metadata_bucket, independent_snapshot, is_expected_client_error, @@ -1812,14 +1813,16 @@ def _get_cost_breakdown_from_logging_obj( def _header_value(value: float | None) -> str | None: - return str(value) if value is not None else None + return str(value) if finite_cost(value) is not None else None def _response_cost_header_value(hidden_params: Mapping[str, object]) -> float | str: if hidden_params.get("cache_hit") is True: return 0.0 response_cost: Final = hidden_params.get("response_cost") - return response_cost if isinstance(response_cost, (int, float, str)) and response_cost else "" + if isinstance(response_cost, str): + return response_cost + return response_cost if isinstance(response_cost, int | float) and finite_cost(response_cost) else "" def cost_headers( @@ -4509,7 +4512,7 @@ class ProxyBaseLLMRequestProcessing: """ if isinstance(chunk, dict): # Use safe_dumps for proper JSON serialization with circular reference detection - chunk_str: Final = safe_dumps(chunk) + chunk_str: Final = safe_dumps(chunk, non_finite_as_null=True) return f"{STREAM_SSE_DATA_PREFIX}{chunk_str}\n\n" else: return chunk @@ -4885,7 +4888,11 @@ class ProxyBaseLLMRequestProcessing: obj, model_name, litellm_logging_obj ) if maybe_modified is not None: - lines[idx] = "data: " + safe_dumps(maybe_modified) + ("\r" if ln.endswith("\r") else "") + lines[idx] = ( + "data: " + + safe_dumps(maybe_modified, non_finite_as_null=True) + + ("\r" if ln.endswith("\r") else "") + ) return "\n".join(lines) return None except Exception: diff --git a/litellm/proxy/db/db_spend_update_writer.py b/litellm/proxy/db/db_spend_update_writer.py index fe804a82e96..d9fb236632b 100644 --- a/litellm/proxy/db/db_spend_update_writer.py +++ b/litellm/proxy/db/db_spend_update_writer.py @@ -34,6 +34,7 @@ from litellm.constants import ( DB_SPEND_UPDATE_JOB_NAME, INTERNAL_CALL_ORIGIN_METADATA_KEY, ) +from litellm.litellm_core_utils.core_helpers import finite_cost, single_log_line from litellm.litellm_core_utils.litellm_logging import coerce_model_access_groups from litellm.litellm_core_utils.safe_json_loads import safe_json_loads from litellm.proxy._types import ( @@ -652,7 +653,14 @@ class DBSpendUpdateWriter: end_time=end_time, llm_router=llm_router, ) - payload["spend"] = response_cost or 0.0 + billable_cost: Final = finite_cost(response_cost) + if response_cost is not None and billable_cost is None: + verbose_proxy_logger.warning( + "Spend tracking - recording a non-finite response_cost=%s as 0 for request_id=%s", + response_cost, + single_log_line(payload["request_id"]), + ) + payload["spend"] = billable_cost or 0.0 if isinstance(payload["startTime"], datetime): payload["startTime"] = payload["startTime"].isoformat() if isinstance(payload["endTime"], datetime): @@ -702,7 +710,7 @@ class DBSpendUpdateWriter: spawn_logging_task( asyncio.get_running_loop(), self._batch_database_updates( - response_cost=response_cost, + response_cost=billable_cost, user_id=user_id, hashed_token=hashed_token, team_id=team_id, diff --git a/litellm/proxy/db/db_transaction_queue/redis_update_buffer.py b/litellm/proxy/db/db_transaction_queue/redis_update_buffer.py index 3a4375ac54a..459f95972ce 100644 --- a/litellm/proxy/db/db_transaction_queue/redis_update_buffer.py +++ b/litellm/proxy/db/db_transaction_queue/redis_update_buffer.py @@ -8,7 +8,7 @@ import asyncio import json from collections.abc import Mapping, Sequence from datetime import datetime -from functools import reduce +from functools import partial, reduce from itertools import chain from typing import TYPE_CHECKING, Any, Final, Literal, TypeAlias, TypeVar, cast @@ -109,6 +109,16 @@ _ValueT = TypeVar("_ValueT") _DECODED_JSON: Final = TypeAdapter(object) +def _zero_for_non_finite_constant(constant: str) -> float: + verbose_proxy_logger.warning( + "Redis spend buffer - reading a parked %s spend as 0, the spend it held is lost", constant + ) + return 0.0 + + +_load_buffered_transaction: Final = partial(json.loads, parse_constant=_zero_for_non_finite_constant) + + def _spend_log_json_default(value: object) -> str: return value.isoformat() if isinstance(value, datetime) else str(value) @@ -118,7 +128,7 @@ def _encode_spend_log_row(row: SpendLogRow) -> str: def _decode_spend_log_row(encoded: str) -> dict[str, object] | None: - decoded: Final = _DECODED_JSON.validate_python(json.loads(encoded)) + decoded: Final = _DECODED_JSON.validate_python(_load_buffered_transaction(encoded)) return decoded if isinstance(decoded, dict) else None @@ -825,13 +835,13 @@ class RedisUpdateBuffer: if slot_result is None: daily_results.append(None) else: - list_of_daily = [json.loads(t) for t in slot_result] + list_of_daily = [_load_buffered_transaction(t) for t in slot_result] aggregated = DailySpendUpdateQueue.get_aggregated_daily_spend_update_transactions(list_of_daily) daily_results.append(aggregated) window_spend: Final = ( WindowSpendUpdateQueue.get_aggregated_window_spend_transactions( - tuple(json.loads(transaction) for transaction in raw_results[6]) + tuple(_load_buffered_transaction(transaction) for transaction in raw_results[6]) ) if raw_results[6] is not None else None @@ -880,7 +890,7 @@ class RedisUpdateBuffer: ) if list_of_transactions is None: return None - return [json.loads(transaction) for transaction in list_of_transactions] + return [_load_buffered_transaction(transaction) for transaction in list_of_transactions] async def get_all_daily_spend_update_transactions_from_redis_buffer( self, @@ -1010,9 +1020,9 @@ class RedisUpdateBuffer: Parses the list of transactions from Redis """ if isinstance(list_of_transactions, list): - return [json.loads(transaction) for transaction in list_of_transactions] + return [_load_buffered_transaction(transaction) for transaction in list_of_transactions] else: - return [json.loads(list_of_transactions)] + return [_load_buffered_transaction(list_of_transactions)] @staticmethod def _combine_list_of_transactions( diff --git a/litellm/proxy/hooks/proxy_track_cost_callback.py b/litellm/proxy/hooks/proxy_track_cost_callback.py index 9c4a4bc775b..0d333d4fd08 100644 --- a/litellm/proxy/hooks/proxy_track_cost_callback.py +++ b/litellm/proxy/hooks/proxy_track_cost_callback.py @@ -16,6 +16,7 @@ from litellm.constants import BACKGROUND_INTERACTION_COST_POLLING_ENABLED, BILLE from litellm.integrations.custom_logger import CustomLogger from litellm.litellm_core_utils.core_helpers import ( budget_reservation_from_metadata, + finite_cost, get_litellm_metadata_from_kwargs, get_metadata_variable_name_from_kwargs, get_parent_otel_span_from_kwargs, @@ -703,7 +704,7 @@ class _ProxyDBLogger(CustomLogger): raw_cost: Final = _str_keyed_fields( sl_object if sl_object is not None else kwargs # pyright: ignore[reportUnknownArgumentType] # untyped success-callback payload ).get("response_cost") - priced_cost: Final = raw_cost if isinstance(raw_cost, (int, float)) else None + priced_cost: Final = finite_cost(raw_cost) response_cost: Final = ( await apply_search_free_tier( kwargs=kwargs, # pyright: ignore[reportUnknownArgumentType] # untyped success-callback payload diff --git a/litellm/proxy/spend_tracking/savings.py b/litellm/proxy/spend_tracking/savings.py index 28602fbbefc..8e79ea5255f 100644 --- a/litellm/proxy/spend_tracking/savings.py +++ b/litellm/proxy/spend_tracking/savings.py @@ -689,9 +689,13 @@ def compute_savings_spend( cost_breakdown=cost_breakdown, ) ) - return SavingsSpend( + savings: Final = SavingsSpend( compression=compression, prompt_caching=prompt_caching, autorouter=0.0 if autorouter is None else autorouter, gateway_injected_caching=gateway_injected_caching, ) + if all(isfinite(value) for value in savings): + return savings + verbose_proxy_logger.warning("Spend tracking - recording non-finite savings %s as 0 for model=%s", savings, model) + return SavingsSpend(*(value if isfinite(value) else 0.0 for value in savings)) diff --git a/litellm/proxy/spend_tracking/spend_tracking_utils.py b/litellm/proxy/spend_tracking/spend_tracking_utils.py index 09b7ce7ee9a..f8d5cba4229 100644 --- a/litellm/proxy/spend_tracking/spend_tracking_utils.py +++ b/litellm/proxy/spend_tracking/spend_tracking_utils.py @@ -912,7 +912,7 @@ def get_logging_payload( user=metadata.get("user_api_key_user_id", "") or "", team_id=metadata.get("user_api_key_team_id", "") or "", organization_id=metadata.get("user_api_key_org_id") or "", - metadata=safe_dumps(clean_metadata), + metadata=safe_dumps(clean_metadata, non_finite_as_null=True), cache_key=cache_key, spend=kwargs.get("response_cost", 0), total_tokens=usage.get("total_tokens", standard_logging_total_tokens), @@ -1224,7 +1224,7 @@ def _get_messages_for_spend_logs_payload( messages: Final = standard_logging_payload.get("messages") if messages is not None: try: - return safe_dumps(messages) + return safe_dumps(messages, non_finite_as_null=True) except Exception: return "{}" return "{}" @@ -1919,7 +1919,8 @@ def _get_proxy_server_request_for_spend_logs_payload( _request_body_json_str: Final = safe_dumps( _stored_request_without_requester_ip_address(stored_request_body) if drop_requester_ip_address - else stored_request_body + else stored_request_body, + non_finite_as_null=True, ) if LITELLM_TRUNCATED_PAYLOAD_FIELD in _request_body_json_str: verbose_proxy_logger.info( @@ -2000,7 +2001,7 @@ def _get_response_for_spend_logs_payload( if isinstance(sanitized_response, str): result_str = strip_null_bytes(sanitized_response) else: - result_str = safe_dumps(sanitized_response) + result_str = safe_dumps(sanitized_response, non_finite_as_null=True) if LITELLM_TRUNCATED_PAYLOAD_FIELD in result_str: verbose_proxy_logger.info( "Spend Log: response was truncated before storing in DB. %s", @@ -2071,7 +2072,7 @@ def spend_log_row_with_retained_metadata( return row metadata: Final = _SPEND_LOGS_METADATA_ADAPTER.validate_json(metadata_json) retained: Final = {name: value for name, value in metadata.items() if fields.keeps(name)} - return {**row, "metadata": safe_dumps(retained)} + return {**row, "metadata": safe_dumps(retained, non_finite_as_null=True)} def _get_status_for_spend_log( diff --git a/litellm/types/llms/openai.py b/litellm/types/llms/openai.py index d9d86cf54b3..a54d66c4beb 100644 --- a/litellm/types/llms/openai.py +++ b/litellm/types/llms/openai.py @@ -1,4 +1,5 @@ import builtins +import math from collections.abc import Iterable, Mapping, Sequence from enum import Enum from os import PathLike @@ -83,6 +84,7 @@ from typing_extensions import ( override, ) +from litellm._logging import verbose_logger from litellm.types.llms.base import BaseLiteLLMOpenAIResponseObject, LiteLLMBaseModel from litellm.types.responses.main import ( CustomToolCallOutputItem, @@ -1446,9 +1448,11 @@ class ResponseAPIUsage(BaseLiteLLMOpenAIResponseObject): @classmethod def parse_cost(cls, v: object) -> object: """Normalise cost: accept either a float or a dict with a ``total_cost`` key.""" - if isinstance(v, dict): - return v.get("total_cost") - return v + cost: Final = v.get("total_cost") if isinstance(v, dict) else v + if isinstance(cost, float) and not math.isfinite(cost): + verbose_logger.warning("Dropping a non-finite provider usage cost") + return None + return cost model_config = ConfigDict(extra="allow") diff --git a/litellm/types/utils.py b/litellm/types/utils.py index 780c9ef6824..55aaf606175 100644 --- a/litellm/types/utils.py +++ b/litellm/types/utils.py @@ -1,4 +1,5 @@ import json +import math import re import time from collections.abc import Collection, Mapping, Sequence @@ -1971,6 +1972,19 @@ class ServerToolUse(LiteLLMBaseModel): return getattr(self, key) +_COST_BREAKDOWN_ADAPTER: Final = TypeAdapter(Mapping[object, object]) + + +def _is_non_finite_float(value: object) -> bool: + return isinstance(value, float) and not math.isfinite(value) + + +def _carries_non_finite_number(cost: object) -> bool: + if not isinstance(cost, Mapping): + return _is_non_finite_float(cost) + return any(_is_non_finite_float(member) for member in _COST_BREAKDOWN_ADAPTER.validate_python(cost).values()) + + class Usage(SafeAttributeModel, CompletionUsage): _cache_creation_input_tokens: int = PrivateAttr( 0 @@ -1978,6 +1992,7 @@ class Usage(SafeAttributeModel, CompletionUsage): _cache_read_input_tokens: int = PrivateAttr( 0 ) # hidden param for prompt caching. Might change, once openai introduces their equivalent. + _dropped_non_finite_cost: bool = PrivateAttr(False) server_tool_use: ServerToolUse | None = None cost: float | None = None @@ -2084,10 +2099,14 @@ class Usage(SafeAttributeModel, CompletionUsage): else: # maintain openai compatibility in usage object if possible del self.server_tool_use - if cost is not None: - self.cost = cost - else: + dropped_non_finite_cost: Final = cost is not None and _carries_non_finite_number(cost) + if cost is None: del self.cost + elif dropped_non_finite_cost: + verbose_logger.warning("Dropping a non-finite provider usage cost") + del self.cost + else: + self.cost = cost ## ANTHROPIC MAPPING ## if "cache_creation_input_tokens" in params and isinstance(params["cache_creation_input_tokens"], int): @@ -2102,6 +2121,11 @@ class Usage(SafeAttributeModel, CompletionUsage): for k, v in params.items(): setattr(self, k, v) + self._dropped_non_finite_cost = dropped_non_finite_cost + + @staticmethod + def dropped_non_finite_cost(usage: "Usage") -> bool: + return usage._dropped_non_finite_cost def __contains__(self, key) -> bool: # Define custom behavior for the 'in' operator diff --git a/tests/integration/spend/test_non_finite_cost_spend_writes.py b/tests/integration/spend/test_non_finite_cost_spend_writes.py new file mode 100644 index 00000000000..9a4d2dd45a4 --- /dev/null +++ b/tests/integration/spend/test_non_finite_cost_spend_writes.py @@ -0,0 +1,166 @@ +import json +from hashlib import sha256 +from pathlib import Path +from typing import Final, Literal +from uuid import uuid4 + +import pytest +from integration._support.client import Gateway, eventually +from integration._support.database import read_rows +from integration._support.openai_wire import answering_model_discovery +from integration._support.process import owned_proxy +from integration._support.wire import Reply, Request, wire_server +from pydantic import JsonValue + +pytestmark: Final = pytest.mark.shares_worker_state + +PROMPT_TOKENS: Final = 5 +COMPLETION_TOKENS: Final = 3 +INPUT_RATE: Final = 0.001 +OUTPUT_RATE: Final = 0.002 +TOKEN_PRICE: Final = PROMPT_TOKENS * INPUT_RATE + COMPLETION_TOKENS * OUTPUT_RATE +OVERFLOWING_INPUT_RATE: Final = 1e308 +POISONED_UPSTREAM: Final = "poisoned-upstream" +CONTROL_UPSTREAM: Final = "control-upstream" +DAILY_TOTALS: Final = "SELECT model_group, SUM(spend)::float8 AS spend, SUM(api_requests)::int8 AS api_requests" +DAILY_USER_QUERY: Final = f'{DAILY_TOTALS} FROM "LiteLLM_DailyUserSpend" WHERE api_key=%s GROUP BY model_group' +DAILY_TEAM_QUERY: Final = f'{DAILY_TOTALS} FROM "LiteLLM_DailyTeamSpend" WHERE api_key=%s GROUP BY model_group' +DAILY_TAG_QUERY: Final = f'{DAILY_TOTALS} FROM "LiteLLM_DailyTagSpend" WHERE api_key=%s AND tag=%s GROUP BY model_group' + +Poison = Literal["reported_nan_cost", "overflowing_price"] + + +def _frame(payload: dict[str, JsonValue]) -> bytes: + return b"data: " + json.dumps(payload).encode() + b"\n\n" + + +def _stream(upstream_model: str, reported_cost: float | None) -> tuple[bytes, ...]: + chunk: Final = {"id": f"chatcmpl-{uuid4().hex}", "object": "chat.completion.chunk", "created": 1} + usage: Final[dict[str, JsonValue]] = { + "prompt_tokens": PROMPT_TOKENS, + "completion_tokens": COMPLETION_TOKENS, + "total_tokens": PROMPT_TOKENS + COMPLETION_TOKENS, + } + reported_usage: Final = usage if reported_cost is None else {**usage, "cost": reported_cost} + return ( + _frame({**chunk, "model": upstream_model, "choices": [{"index": 0, "delta": {"content": "ok"}}]}), + _frame( + { + **chunk, + "model": upstream_model, + "choices": [{"index": 0, "delta": {}, "finish_reason": "stop"}], + "usage": reported_usage, + } + ), + b"data: [DONE]\n\n", + ) + + +def _deployment(alias: str, upstream_model: str, wire_url: str, input_rate: float) -> dict[str, JsonValue]: + return { + "model_name": alias, + "litellm_params": { + "model": f"openai/{upstream_model}", + "api_base": f"{wire_url}/v1", + "api_key": "integration-provider-key", + "input_cost_per_token": input_rate, + "output_cost_per_token": OUTPUT_RATE, + }, + } + + +def _proxy_config(directory: Path, deployments: tuple[dict[str, JsonValue], ...]) -> Path: + config: Final = directory / "non_finite_cost_config.yaml" + config.write_text( + json.dumps( + { + "model_list": list(deployments), + "general_settings": { + "master_key": "os.environ/LITELLM_MASTER_KEY", + "database_url": "os.environ/DATABASE_URL", + "disable_spend_logs": False, + "proxy_batch_write_at": 1, + "proxy_batch_polling_interval": 1, + }, + "router_settings": {"disable_cooldowns": True}, + } + ) + ) + return config + + +def _by_model_group(rows: list[dict[str, JsonValue]]) -> dict[str, float]: + return {str(row["model_group"]): float(str(row["spend"])) for row in rows} + + +def _both_requests_rows(query: str, parameters: tuple[str, ...]) -> list[dict[str, JsonValue]]: + return eventually(lambda: read_rows(query, parameters), lambda rows: len(rows) == 2, seconds=70) + + +@pytest.mark.parametrize("poison", ["reported_nan_cost", "overflowing_price"]) +@pytest.mark.timeout(240) +def test_a_non_finite_cost_never_blocks_the_next_requests_spend_rows( + gateway: Gateway, tmp_path: Path, poison: Poison +) -> None: + poisoned_alias: Final = f"non-finite-cost-{uuid4().hex}" + control_alias: Final = f"finite-cost-{uuid4().hex}" + tag: Final = f"non-finite-cost-{uuid4().hex}" + poisoned_spend: Final = TOKEN_PRICE if poison == "reported_nan_cost" else 0.0 + poisoned_input_rate: Final = INPUT_RATE if poison == "reported_nan_cost" else OVERFLOWING_INPUT_RATE + + def respond(request: Request) -> Reply: + assert request.target.endswith("/chat/completions"), request.target + upstream_model: Final = json.loads(request.body)["model"] + reported_cost: Final = ( + float("nan") if upstream_model == POISONED_UPSTREAM and poison == "reported_nan_cost" else None + ) + return Reply(content_type="text/event-stream", chunks=_stream(upstream_model, reported_cost)) + + with wire_server(answering_model_discovery(respond)) as wire: + config: Final = _proxy_config( + tmp_path, + ( + _deployment(poisoned_alias, POISONED_UPSTREAM, wire.url, poisoned_input_rate), + _deployment(control_alias, CONTROL_UPSTREAM, wire.url, INPUT_RATE), + ), + ) + with owned_proxy(gateway, tmp_path, {}, config=config) as candidate, candidate.scenario() as scenario: + team: Final = scenario.team() + key: Final = scenario.key(user_id=f"non-finite-cost-{uuid4().hex}", team_id=team) + digest: Final = sha256(key.encode()).hexdigest() + responses: Final = tuple( + candidate.request( + "POST", + "/v1/chat/completions", + { + "model": alias, + "messages": [{"role": "user", "content": f"non-finite cost {uuid4().hex}"}], + "stream": True, + "metadata": {"tags": [tag]}, + }, + key=key, + ) + for alias in (poisoned_alias, control_alias) + ) + assert all(response.status_code == 200 for response in responses), [response.text for response in responses] + + expected: Final = {poisoned_alias: pytest.approx(poisoned_spend), control_alias: pytest.approx(TOKEN_PRICE)} + spend_logs: Final = _both_requests_rows( + 'SELECT model_group, spend FROM "LiteLLM_SpendLogs" WHERE api_key=%s', (digest,) + ) + assert _by_model_group(spend_logs) == expected, spend_logs + daily_rows: Final = ( + _both_requests_rows(DAILY_USER_QUERY, (digest,)), + _both_requests_rows(DAILY_TEAM_QUERY, (digest,)), + _both_requests_rows(DAILY_TAG_QUERY, (digest, tag)), + ) + assert [_by_model_group(rows) for rows in daily_rows] == [expected] * 3, daily_rows + assert [{row["api_requests"] for row in rows} for rows in daily_rows] == [{1}] * 3, daily_rows + key_total: Final = poisoned_spend + TOKEN_PRICE + key_spend: Final = eventually( + lambda: read_rows('SELECT spend FROM "LiteLLM_VerificationToken" WHERE token=%s', (digest,)), + lambda rows: len(rows) == 1 and float(str(rows[0]["spend"])) == pytest.approx(key_total), + seconds=70, + return_last_on_timeout=True, + ) + assert float(str(key_spend[0]["spend"])) == pytest.approx(key_total), key_spend diff --git a/tests/unit/litellm_core_utils/llm_response_utils/test_convert_dict_to_response.py b/tests/unit/litellm_core_utils/llm_response_utils/test_convert_dict_to_response.py index de8d6a7a8c5..d0f5bfc6b9f 100644 --- a/tests/unit/litellm_core_utils/llm_response_utils/test_convert_dict_to_response.py +++ b/tests/unit/litellm_core_utils/llm_response_utils/test_convert_dict_to_response.py @@ -3089,6 +3089,50 @@ def test_convert_to_model_response_object_function_output(): assert result._hidden_params is not None +@pytest.mark.parametrize( + "reported_cost", + [ + float("nan"), + float("inf"), + {"input_tokens_cost": 0.0001, "request_cost": 0.005, "total_cost": float("nan")}, + {"input_tokens_cost": float("inf"), "request_cost": 0.005, "total_cost": 0.0051}, + ], + ids=["nan", "inf", "breakdown-with-nan-total", "breakdown-with-inf-part"], +) +def test_a_non_finite_provider_usage_cost_is_dropped_so_the_response_is_strict_json(reported_cost: object): + response: Final = convert_to_model_response_object( + response_object={ + "id": "chatcmpl-non-finite-cost", + "object": "chat.completion", + "created": 1, + "model": "gpt-4o", + "choices": [{"index": 0, "message": {"role": "assistant", "content": "ok"}, "finish_reason": "stop"}], + "usage": {"prompt_tokens": 3, "completion_tokens": 1, "total_tokens": 4, "cost": reported_cost}, + }, + model_response_object=ModelResponse(), + ) + + usage: Final = json.loads(json.dumps(response.model_dump(), allow_nan=False))["usage"] + assert "cost" not in usage + assert usage["total_tokens"] == 4 + + +def test_a_finite_provider_cost_breakdown_with_non_string_keys_is_kept(): + response: Final = convert_to_model_response_object( + response_object={ + "id": "chatcmpl-int-key-cost", + "object": "chat.completion", + "created": 1, + "model": "gpt-4o", + "choices": [{"index": 0, "message": {"role": "assistant", "content": "ok"}, "finish_reason": "stop"}], + "usage": {"prompt_tokens": 3, "completion_tokens": 1, "total_tokens": 4, "cost": {1: 0.5}}, + }, + model_response_object=ModelResponse(), + ) + + assert response.usage.cost == {1: 0.5} + + _URL_CITATION: Final = { "type": "url_citation", "url_citation": { diff --git a/tests/unit/litellm_core_utils/test_core_helpers.py b/tests/unit/litellm_core_utils/test_core_helpers.py index 1c86c4e30d8..93defc0f04f 100644 --- a/tests/unit/litellm_core_utils/test_core_helpers.py +++ b/tests/unit/litellm_core_utils/test_core_helpers.py @@ -26,15 +26,24 @@ from litellm.litellm_core_utils.core_helpers import ( normalize_drop_params, preserve_upstream_non_openai_attributes, process_response_headers, + provider_reported_usage_cost, qualify_agentic_followup_model, qualify_provider_stripped_model, reconstruct_model_name, redact_nested_match_and_regex_keys, + response_cost_in_hidden_params, set_provider_response_headers_in_hidden_params, unbind_budget_reservation_from_callbacks, ) from litellm.proxy._types import UserAPIKeyAuth -from litellm.types.utils import Delta, ImageResponse, ModelResponseStream, StreamingChoices, TranscriptionResponse +from litellm.types.utils import ( + Delta, + ImageResponse, + ModelResponse, + ModelResponseStream, + StreamingChoices, + TranscriptionResponse, +) def test_parent_otel_span_reads_dict_like_metadata_by_index(): @@ -790,6 +799,23 @@ class TestReceiveStampSurvivesASecondMetadataBucket: assert get_litellm_metadata_from_kwargs(kwargs)["litellm_received_at"] is own +@pytest.mark.parametrize("reported", [float("nan"), float("inf"), {"total_cost": float("nan")}]) +def test_a_non_finite_provider_reported_cost_counts_as_unreported(reported: object) -> None: + assert provider_reported_usage_cost({"cost": reported}, "openrouter") is None + assert provider_reported_usage_cost({"cost": {"total_cost": 0.25}}, "openrouter") == 0.25 + + +@pytest.mark.parametrize("header_cost", [float("nan"), float("inf")]) +def test_a_non_finite_response_cost_header_counts_as_unreported(header_cost: float) -> None: + response: Final = ModelResponse() + response._hidden_params = {"additional_headers": {RESPONSE_COST_HEADER: header_cost}} + finite: Final = ModelResponse() + finite._hidden_params = {"additional_headers": {RESPONSE_COST_HEADER: 0.25}} + + assert response_cost_in_hidden_params(response) is None + assert response_cost_in_hidden_params(finite) == 0.25 + + def test_billed_usage_cost_bills_any_chat_upstream_without_its_own_reader() -> None: from types import SimpleNamespace diff --git a/tests/unit/litellm_core_utils/test_litellm_logging.py b/tests/unit/litellm_core_utils/test_litellm_logging.py index b3ae3bf6240..336e9f7d24a 100644 --- a/tests/unit/litellm_core_utils/test_litellm_logging.py +++ b/tests/unit/litellm_core_utils/test_litellm_logging.py @@ -6308,6 +6308,60 @@ def test_standard_logging_payload_resolves_kubernetes_pod_routing_from_metadata_ assert payload["metadata"][KUBERNETES_POD_ROUTING_KEY] == expected +@pytest.mark.parametrize("response_cost", [float("nan"), float("inf"), float("-inf")]) +def test_get_standard_logging_object_payload_logs_a_non_finite_response_cost_as_zero( + logging_obj: LitellmLogging, response_cost: float +) -> None: + from datetime import datetime + + from litellm.litellm_core_utils.litellm_logging import get_standard_logging_object_payload + + now: Final = datetime(2026, 1, 1) + payload: Final = get_standard_logging_object_payload( + kwargs={"model": "gpt-4o", "messages": [], "litellm_params": {}, "response_cost": response_cost}, + init_response_obj={}, + start_time=now, + end_time=now, + logging_obj=logging_obj, + status="success", + ) + + assert payload is not None + assert payload["response_cost"] == 0 + + +def test_an_overflowing_custom_price_records_a_zero_response_cost(caplog: pytest.LogCaptureFixture) -> None: + messages: Final = [{"role": "user", "content": "hello world"}] + start_time: Final = datetime.datetime.now() + logging_obj: Final = Logging( + model="gpt-4o", + messages=messages, + stream=False, + call_type="completion", + start_time=start_time, + litellm_call_id="overflow\nforged log line", + function_id="overflow", + ) + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + response: Final = litellm.completion( + model="gpt-4o", + mock_response="hi there", + messages=messages, + input_cost_per_token=1e308, + output_cost_per_token=1e308, + litellm_logging_obj=logging_obj, + ) + logging_obj._process_hidden_params_and_response_cost( + response, start_time=start_time, end_time=datetime.datetime.now() + ) + + assert response._hidden_params["response_cost"] == 0.0 + payload: Final = logging_obj.model_call_details["standard_logging_object"] + assert json.loads(json.dumps(payload, allow_nan=False, default=str))["response_cost"] == 0.0 + warnings: Final = [r.getMessage() for r in caplog.records if "non-finite response_cost" in r.getMessage()] + assert warnings == ["Recording a non-finite response_cost as 0 for litellm_call_id=overflowforged log line"] + + def test_get_standard_logging_object_payload_carries_matched_access_groups(logging_obj): """Access groups stamped at auth time reach the logging payload, so integrations see what a request billed.""" from datetime import datetime diff --git a/tests/unit/litellm_core_utils/test_safe_json_dumps.py b/tests/unit/litellm_core_utils/test_safe_json_dumps.py index 1f2664f33cb..1b8bdd21326 100644 --- a/tests/unit/litellm_core_utils/test_safe_json_dumps.py +++ b/tests/unit/litellm_core_utils/test_safe_json_dumps.py @@ -169,6 +169,31 @@ def test_null_byte_stripped_in_fallback_str(): assert json.loads(out)["obj"] == "objrepr" +@pytest.mark.parametrize("value", [float("nan"), float("inf"), float("-inf")]) +def test_non_finite_floats_are_written_as_json_null(value: float) -> None: + out = safe_dumps( + {"cost_breakdown": {"total_cost": value, "output_cost": 0.006}, "costs": [value]}, non_finite_as_null=True + ) + + assert json.loads(out, parse_constant=lambda token: pytest.fail(f"non-JSON token {token}")) == { + "cost_breakdown": {"total_cost": None, "output_cost": 0.006}, + "costs": [None], + } + + +def test_a_set_holding_a_non_finite_float_sorts_its_finite_members_and_puts_null_last() -> None: + assert json.loads(safe_dumps({"costs": {2.0, float("inf"), 1.0}}, non_finite_as_null=True)) == { + "costs": [1.0, 2.0, None] + } + + +def test_non_finite_floats_keep_their_value_unless_the_caller_asks_for_null() -> None: + assert json.loads(safe_dumps({"timeout": float("inf"), "costs": [float("-inf")]})) == { + "timeout": float("inf"), + "costs": [float("-inf")], + } + + def test_clean_strings_are_not_run_through_replace(): """Regression for LIT-3910. diff --git a/tests/unit/litellm_core_utils/test_streaming_chunk_builder_utils.py b/tests/unit/litellm_core_utils/test_streaming_chunk_builder_utils.py index 28587bf4749..c8fa9c94c17 100644 --- a/tests/unit/litellm_core_utils/test_streaming_chunk_builder_utils.py +++ b/tests/unit/litellm_core_utils/test_streaming_chunk_builder_utils.py @@ -1077,6 +1077,47 @@ def test_cost_field_in_usage_chunks(): assert usage.completion_tokens == 5 + +def _perplexity_usage_chunk(cost: Mapping[str, float], finish_reason: str | None) -> ModelResponseStream: + return ModelResponseStream( + id="chatcmpl-pplx", + created=1745513206, + model="sonar", + choices=[StreamingChoices(finish_reason=finish_reason, index=0, delta=Delta(content="ok"))], + usage={"prompt_tokens": 26, "completion_tokens": 1, "total_tokens": 27, "cost": dict(cost)}, + ) + + +@pytest.mark.parametrize("poisoned_field", ["request_cost", "total_cost"]) +def test_a_non_finite_final_stream_cost_does_not_fall_back_to_an_earlier_chunk_cost(poisoned_field: str): + final_cost: Final = {"input_tokens_cost": 0.0012, "output_tokens_cost": 1e-05, "request_cost": 0.0025} + chunks: Final = [ + _perplexity_usage_chunk({"input_tokens_cost": 0, "output_tokens_cost": 0, "total_cost": 0}, None), + _perplexity_usage_chunk({**final_cost, "total_cost": 0.00371, poisoned_field: float("nan")}, "stop"), + ] + + response: Final = stream_chunk_builder(chunks=chunks) + + assert response is not None + assert getattr(response.usage, "cost", None) is None + assert response.usage.total_tokens == 27 + + +@pytest.mark.parametrize("provider_key", ["dropped_non_finite_cost", "_dropped_non_finite_cost"]) +def test_a_provider_usage_key_named_like_the_dropped_cost_flag_keeps_the_stream_cost(provider_key: str): + chunk: Final = ModelResponseStream( + id="chatcmpl-flag-key", + created=1745513206, + model="sonar", + choices=[StreamingChoices(finish_reason="stop", index=0, delta=Delta(content="ok"))], + usage={"prompt_tokens": 26, "completion_tokens": 1, "total_tokens": 27, "cost": 0.00371, provider_key: True}, + ) + + response: Final = stream_chunk_builder(chunks=[chunk]) + + assert response is not None + assert getattr(response.usage, "cost", None) == 0.00371 + def test_stream_chunk_builder_tolerates_trailing_chunk_without_choices(): """Regression for https://github.com/BerriAI/litellm/issues/32051 diff --git a/tests/unit/llms/openrouter/chat/test_openrouter_chat_transformation.py b/tests/unit/llms/openrouter/chat/test_openrouter_chat_transformation.py index 59355dfabcb..888c47505a4 100644 --- a/tests/unit/llms/openrouter/chat/test_openrouter_chat_transformation.py +++ b/tests/unit/llms/openrouter/chat/test_openrouter_chat_transformation.py @@ -1,6 +1,7 @@ import json from typing import Final +from unittest.mock import MagicMock import httpx import pytest @@ -464,6 +465,34 @@ def test_openrouter_cost_tracking_non_streaming(): ) + +@pytest.mark.parametrize("cost", [float("nan"), float("inf")]) +def test_openrouter_chat_leaves_a_non_finite_provider_cost_out_of_the_response_headers(cost: float) -> None: + body: Final = { + "id": "gen-123", + "object": "chat.completion", + "created": 1234567890, + "model": "openai/gpt-5.6-luna", + "choices": [{"message": {"role": "assistant", "content": "ok"}, "finish_reason": "stop", "index": 0}], + "usage": {"prompt_tokens": 10, "completion_tokens": 1, "total_tokens": 11, "cost": cost}, + } + + result: Final = OpenrouterConfig().transform_response( + model="openrouter/openai/gpt-5.6-luna", + raw_response=httpx.Response( + 200, content=json.dumps(body).encode(), headers={"content-type": "application/json"} + ), + model_response=litellm.ModelResponse(), + logging_obj=MagicMock(), + request_data={}, + messages=[{"role": "user", "content": "Reply with ok"}], + optional_params={}, + litellm_params={}, + encoding=None, + ) + + assert "llm_provider-x-litellm-response-cost" not in result._hidden_params.get("additional_headers", {}) + def test_openrouter_cost_tracking_streaming(): """ Test OpenRouter cost tracking for streaming completions. diff --git a/tests/unit/llms/openrouter/responses/test_openrouter_responses_transformation.py b/tests/unit/llms/openrouter/responses/test_openrouter_responses_transformation.py index 19517e58b31..78d7161c388 100644 --- a/tests/unit/llms/openrouter/responses/test_openrouter_responses_transformation.py +++ b/tests/unit/llms/openrouter/responses/test_openrouter_responses_transformation.py @@ -9,6 +9,8 @@ reasoning.encrypted_content for multi-turn stateless workflows. Related issue: https://github.com/BerriAI/litellm/issues/22189 """ +import json +from typing import Final from unittest.mock import MagicMock import httpx @@ -157,6 +159,29 @@ def test_openrouter_responses_returns_provider_reported_cost(cost): assert get_response_cost_from_hidden_params(response.hidden_params) == cost +@pytest.mark.parametrize("cost", [float("nan"), float("inf")]) +def test_openrouter_responses_drops_a_non_finite_provider_cost(cost: float) -> None: + body: Final = { + "id": "resp_test", + "created_at": 1, + "model": "unmapped-model", + "object": "response", + "output": [], + "status": "completed", + "usage": {"input_tokens": 2, "output_tokens": 3, "total_tokens": 5, "cost": cost}, + } + response: Final = OpenRouterResponsesAPIConfig().transform_response_api_response( + model="openrouter/unmapped-model", + raw_response=httpx.Response( + 200, content=json.dumps(body).encode(), headers={"content-type": "application/json"} + ), + logging_obj=MagicMock(), + ) + + assert get_response_cost_from_hidden_params(response.hidden_params) is None + assert json.loads(json.dumps(response.model_dump(), allow_nan=False, default=str))["usage"]["cost"] is None + + @pytest.mark.parametrize( "event_type", ["response.completed", "response.incomplete", "response.failed"], diff --git a/tests/unit/llms/perplexity/test_perplexity_cost_calculator.py b/tests/unit/llms/perplexity/test_perplexity_cost_calculator.py index 83c71479311..380f85445e5 100644 --- a/tests/unit/llms/perplexity/test_perplexity_cost_calculator.py +++ b/tests/unit/llms/perplexity/test_perplexity_cost_calculator.py @@ -6,9 +6,11 @@ search queries, and reasoning tokens. """ import json +import logging import math import os from datetime import datetime, timezone +from typing import Final from unittest.mock import patch import pytest @@ -140,6 +142,26 @@ class TestPerplexityCostCalculator: assert prompt_cost == 0.0 assert completion_cost == 0.008 + @pytest.mark.parametrize( + "reported_cost", + [float("nan"), float("inf"), {"total_cost": float("nan")}], + ids=["nan", "inf", "nan-total-cost-dict"], + ) + def test_a_non_finite_perplexity_cost_is_priced_like_no_reported_cost_with_a_warning( + self, reported_cost: float | dict, caplog: pytest.LogCaptureFixture + ): + unreported: Final = Usage(prompt_tokens=100, completion_tokens=50, total_tokens=150) + reported: Final = Usage(prompt_tokens=100, completion_tokens=50, total_tokens=150) + reported.cost = reported_cost + + token_priced: Final = perplexity_cost_per_token(model="sonar-pro", usage=unreported) + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + reported_priced: Final = perplexity_cost_per_token(model="sonar-pro", usage=reported) + + assert sum(token_priced) > 0 + assert reported_priced == token_priced + assert [record for record in caplog.records if "non-finite provider-reported cost" in record.getMessage()] + OFF_PEAK_MODEL = "sonar-off-peak-test" OFF_PEAK_WINDOW = "14:00-00:00" INSIDE_WINDOW = datetime(2026, 9, 3, 17, 25, tzinfo=timezone.utc) diff --git a/tests/unit/proxy/db/db_transaction_queue/test_redis_update_buffer.py b/tests/unit/proxy/db/db_transaction_queue/test_redis_update_buffer.py index 24c12e06c9c..65da5630f33 100644 --- a/tests/unit/proxy/db/db_transaction_queue/test_redis_update_buffer.py +++ b/tests/unit/proxy/db/db_transaction_queue/test_redis_update_buffer.py @@ -1,10 +1,12 @@ import json +import logging from datetime import datetime, timezone from typing import Final from unittest.mock import AsyncMock, MagicMock, patch import pytest +from litellm._logging import verbose_proxy_logger from litellm.constants import REDIS_DAILY_TAG_SPEND_UPDATE_BUFFER_KEY from litellm.litellm_core_utils.safe_json_dumps import safe_dumps from litellm.proxy._types import DailyTagSpendTransaction @@ -379,6 +381,104 @@ async def test_org_member_spend_is_summed_across_pods_and_restored_on_rpush_fail assert restored_spend["org_member_list_transactions"] == {member_key: 1.5} +@pytest.mark.asyncio +async def test_a_single_parked_batch_with_a_nan_spend_drains_as_zero( + redis_update_buffer: RedisUpdateBuffer, mock_redis_cache: AsyncMock +): + mock_redis_cache.async_lpop = AsyncMock( + return_value=json.dumps({"key_list_transactions": {"nan-key": float("nan"), "ok-key": 0.5}}, allow_nan=True) + ) + + drained: Final = await redis_update_buffer.get_all_update_transactions_from_redis_buffer() + + assert drained is not None + assert drained["key_list_transactions"] == {"nan-key": 0.0, "ok-key": 0.5} + + +PARKED_DAILY_KEY: Final = "user-1_2025-01-15_api-key-123_gpt-4o_openai_/chat/completions" + + +def _parked_daily_batch(spend: float) -> str: + return json.dumps( + { + PARKED_DAILY_KEY: { + "spend": spend, + "prompt_tokens": 12, + "completion_tokens": 7, + "api_requests": 1, + "successful_requests": 1, + "failed_requests": 0, + } + }, + allow_nan=True, + ) + + +@pytest.mark.asyncio +async def test_a_nan_spend_parked_by_an_older_version_drains_from_the_pipeline_as_zero( + redis_update_buffer: RedisUpdateBuffer, mock_redis_cache: AsyncMock +): + entity_batch: Final = json.dumps({"key_list_transactions": {"hashed-key": float("nan")}}, allow_nan=True) + window_batches: Final = tuple( + json.dumps( + [ + { + "entity_type": "key", + "entity_id": "hashed-key", + "window_duration": "30d", + "window_start": "2026-08-01T00:00:00.000000", + "spend": spend, + "started_at": None, + } + ], + allow_nan=True, + ) + for spend in (float("nan"), 3.0) + ) + mock_redis_cache.async_lpop_pipeline = AsyncMock( + return_value=[ + [entity_batch], + [_parked_daily_batch(float("nan")), _parked_daily_batch(0.0042)], + None, + None, + None, + None, + list(window_batches), + ] + ) + + ( + db_spend, + daily_user, + *_rest, + window_spend, + ) = await redis_update_buffer.get_all_transactions_from_redis_buffer_pipeline() + + assert db_spend is not None + assert db_spend["key_list_transactions"] == {"hashed-key": 0} + assert daily_user is not None + assert daily_user[PARKED_DAILY_KEY]["spend"] == 0.0042 + assert daily_user[PARKED_DAILY_KEY]["api_requests"] == 2 + assert window_spend is not None + assert [transaction["spend"] for transaction in window_spend] == [3.0] + + +@pytest.mark.asyncio +async def test_a_nan_spend_parked_by_an_older_version_drains_from_a_daily_key_as_zero( + redis_update_buffer: RedisUpdateBuffer, mock_redis_cache: AsyncMock, caplog: pytest.LogCaptureFixture +): + mock_redis_cache.async_lpop = AsyncMock( + return_value=[_parked_daily_batch(float("inf")), _parked_daily_batch(0.0042)] + ) + + with caplog.at_level(logging.WARNING, logger=verbose_proxy_logger.name): + daily_tag: Final = await redis_update_buffer.get_all_daily_tag_spend_update_transactions_from_redis_buffer() + + assert daily_tag is not None + assert daily_tag[PARKED_DAILY_KEY]["spend"] == 0.0042 + assert [record.getMessage() for record in caplog.records if "Infinity" in record.getMessage()] != [] + + @pytest.mark.asyncio async def test_get_all_transactions_from_redis_buffer_pipeline_no_redis(): """When redis_cache is None, should return all Nones""" @@ -724,6 +824,10 @@ class _ListRedis: def __init__(self) -> None: self.rows: list[str] = [] + async def async_rpush(self, key: str, values: list[str]) -> int: + self.rows.extend(values) + return len(self.rows) + async def async_rpush_and_trim(self, key: str, values: list[str], max_len: int) -> int: self.rows.extend(values) pushed_len = len(self.rows) @@ -752,6 +856,32 @@ async def test_store_spend_logs_in_redis_drops_oldest_rows_past_the_cap(): assert await buffer.get_spend_logs_from_redis_buffer(limit=10) == () +@pytest.mark.asyncio +@pytest.mark.parametrize("spend", [float("nan"), float("inf")]) +async def test_a_non_finite_daily_spend_buffered_in_redis_drains_as_zero(spend: float): + buffer: Final = RedisUpdateBuffer(redis_cache=_ListRedis()) + queue: Final = DailySpendUpdateQueue() + await queue.add_update(json.loads(_parked_daily_batch(spend))) + await queue.add_update(json.loads(_parked_daily_batch(0.0042))) + + await buffer.store_in_memory_daily_tag_spend_updates_in_redis(queue) + drained: Final = await buffer.get_all_daily_tag_spend_update_transactions_from_redis_buffer() + + assert drained is not None + assert drained[PARKED_DAILY_KEY]["spend"] == 0.0 + assert drained[PARKED_DAILY_KEY]["api_requests"] == 2 + + +@pytest.mark.asyncio +async def test_a_spend_log_row_parked_with_a_nan_spend_is_read_back_with_zero_spend(): + buffer: Final = RedisUpdateBuffer(redis_cache=_ListRedis()) + buffer.should_commit_spend_updates_to_redis = MagicMock(return_value=True) + + assert await buffer.store_spend_logs_in_redis([{"request_id": "nan-row", "spend": float("nan")}]) is True + + assert await buffer.get_spend_logs_from_redis_buffer(limit=10) == ({"request_id": "nan-row", "spend": 0.0},) + + @pytest.mark.asyncio async def test_store_spend_logs_in_redis_reports_failure_without_redis(): buffer = RedisUpdateBuffer(redis_cache=None) diff --git a/tests/unit/proxy/db/test_db_spend_update_writer.py b/tests/unit/proxy/db/test_db_spend_update_writer.py index c3a93fd89be..298b153f621 100644 --- a/tests/unit/proxy/db/test_db_spend_update_writer.py +++ b/tests/unit/proxy/db/test_db_spend_update_writer.py @@ -2226,6 +2226,60 @@ async def test_shutdown_drain_finishes_the_rollups_update_database_started(): assert (team_row["api_requests"], team_row["spend"]) == (1, 0.1) +@pytest.mark.asyncio +@pytest.mark.parametrize("response_cost", [float("nan"), float("inf"), float("-inf")]) +async def test_update_database_records_a_non_finite_response_cost_as_zero(response_cost: float) -> None: + db_writer: Final = DBSpendUpdateWriter() + prisma: Final = _tool_usage_prisma() + prisma.get_request_status = MagicMock(return_value="success") + response: Final = litellm.ModelResponse( + model="gpt-5.6-luna", + usage=litellm.Usage(prompt_tokens=10, completion_tokens=5, total_tokens=15), + ) + with ( + patch("litellm.proxy.proxy_server.disable_spend_logs", False), + patch("litellm.proxy.proxy_server.prisma_client", prisma), + patch("litellm.proxy.proxy_server.user_api_key_cache", MagicMock()), + patch("litellm.proxy.proxy_server.litellm_proxy_budget_name", "test-budget"), + ): + await db_writer.update_database( + token="test-token", + user_id="test-user", + end_user_id=None, + team_id="test-team", + org_id=None, + kwargs={ + "model": "gpt-5.6-luna", + "custom_llm_provider": "openai", + "call_type": "acompletion", + "litellm_call_id": "req-non-finite-cost", + "response_cost": response_cost, + "litellm_params": {"metadata": {"user_api_key": "hashed-key", "user_api_key_team_id": "test-team"}}, + }, + completion_response=response, + start_time=datetime.now(timezone.utc), + end_time=datetime.now(timezone.utc), + response_cost=response_cost, + ) + await drain_logging_tasks() + + assert [row["spend"] for row in prisma.spend_log_transactions] == [0] + spend: Final = await db_writer.spend_update_queue.flush_and_get_aggregated_db_spend_update_transactions() + assert spend["key_list_transactions"] == {"test-token": 0} + assert spend["team_list_transactions"] == {"test-team": 0} + daily_user: Final = ( + await db_writer.daily_spend_update_queue.flush_and_get_aggregated_daily_spend_update_transactions() + ) + daily_team: Final = ( + await db_writer.daily_team_spend_update_queue.flush_and_get_aggregated_daily_spend_update_transactions() + ) + assert daily_user is not None and daily_team is not None + (user_row,) = daily_user.values() + (team_row,) = daily_team.values() + assert (user_row["api_requests"], user_row["spend"]) == (1, 0) + assert (team_row["api_requests"], team_row["spend"]) == (1, 0) + + @pytest.mark.asyncio async def test_shutdown_drain_waits_for_an_in_flight_rollup_enqueue(): """ diff --git a/tests/unit/proxy/pass_through_endpoints/llm_provider_handlers/test_typesafe_passthrough_logging_handler.py b/tests/unit/proxy/pass_through_endpoints/llm_provider_handlers/test_typesafe_passthrough_logging_handler.py index 4011bc0294a..7f8aac983f5 100644 --- a/tests/unit/proxy/pass_through_endpoints/llm_provider_handlers/test_typesafe_passthrough_logging_handler.py +++ b/tests/unit/proxy/pass_through_endpoints/llm_provider_handlers/test_typesafe_passthrough_logging_handler.py @@ -323,6 +323,35 @@ def test_openrouter_decisions_response_uses_provider_cost_without_cost_map_row( assert response["kwargs"]["combined_usage_object"].total_tokens == 302 +@pytest.mark.parametrize("reported_cost", [float("nan"), float("inf")]) +def test_openrouter_non_finite_cost_logs_the_token_price(monkeypatch: pytest.MonkeyPatch, reported_cost: float) -> None: + monkeypatch.setitem( + litellm.model_cost, + "openrouter/typesafe/jev-1.13", + {"input_cost_per_token": 1.0, "output_cost_per_token": 3.0, "litellm_provider": "openrouter", "mode": "chat"}, + ) + logging_obj: Final = _logging_obj() + + result: Final = TypeSafePassthroughLoggingHandler.typesafe_passthrough_handler( + httpx_response=_response(), + response_body={ + "model": "typesafe/jev-1.13-20260917", + "usage": {"input_tokens": 2, "output_tokens": 1, "cost": reported_cost}, + }, + logging_obj=logging_obj, + url_route="https://openrouter.ai/api/alpha/decisions", + result='{"answers": {}}', + start_time=datetime(2000, 1, 1), + end_time=datetime(2000, 1, 1), + cache_hit=False, + request_body={"model": "typesafe/jev-1.13"}, + custom_llm_provider="openrouter", + ) + + assert result["kwargs"]["response_cost"] == pytest.approx(5.0) + assert logging_obj.model_call_details["response_cost"] == pytest.approx(5.0) + + def test_success_handler_dispatches_openrouter_to_the_shared_handler(): logging_obj = _logging_obj() normalized = PassThroughEndpointLogging().normalize_llm_passthrough_logging_payload( diff --git a/tests/unit/proxy/spend_tracking/test_savings.py b/tests/unit/proxy/spend_tracking/test_savings.py index f9566fd1ba0..8503a124b01 100644 --- a/tests/unit/proxy/spend_tracking/test_savings.py +++ b/tests/unit/proxy/spend_tracking/test_savings.py @@ -137,6 +137,34 @@ def test_compression_savings_priced_at_input_rate(): assert result.prompt_caching == 0.0 +def test_an_overflowing_deployment_price_records_finite_savings() -> None: + router: Final = Router( + model_list=[ + { + "model_name": "overflowing-sonnet", + "litellm_params": { + "model": "anthropic/claude-sonnet-5-5", + "input_cost_per_token": 1e308, + "cache_read_input_token_cost": 1e-07, + }, + }, + ] + ) + deployment_id: Final = router.get_model_list(model_name="overflowing-sonnet")[0]["model_info"]["id"] + + result: Final = compute_savings_spend( + model="claude-sonnet-5-5", + custom_llm_provider="anthropic", + compression_saved_tokens=4389, + gateway_injected_cache=True, + usage_object=_caching_usage(read=1000, written=0), + model_id=deployment_id, + llm_router=lambda: router, + ) + + assert result == (0.0, 0.0, 0.0, 0.0) + + def test_prompt_caching_savings_priced_at_input_minus_cache_read(): input_cost, cache_read_cost = _anthropic_costs("claude-sonnet-5") # A model that supports prompt caching must charge less to read from cache; diff --git a/tests/unit/proxy/test_common_request_processing.py b/tests/unit/proxy/test_common_request_processing.py index 2db8941d304..a7596b51e3e 100644 --- a/tests/unit/proxy/test_common_request_processing.py +++ b/tests/unit/proxy/test_common_request_processing.py @@ -12496,6 +12496,55 @@ class TestResponseCostFromLoggingObjForCacheHits: ) assert {name: headers.get(name) for name in cost_headers} == dict.fromkeys(cost_headers, expected_header) + @pytest.mark.asyncio + @pytest.mark.parametrize( + ("response_cost", "expected_header"), + ((float("nan"), None), (float("inf"), None), (0.0042, "0.0042"), (2, "2")), + ids=("nan", "inf", "finite", "int"), + ) + async def test_a_non_finite_response_cost_sends_no_cost_header( + self, response_cost: float, expected_header: str | None + ): + response: Final = litellm.ModelResponse() + response._hidden_params = {"response_cost": response_cost} + + headers: Final = await ProxyBaseLLMRequestProcessing.build_litellm_proxy_success_headers_from_llm_response( + response=response, + request_data={"model": "gpt-5.4-mini"}, + request=Request({"type": "http", "method": "POST", "headers": []}), + user_api_key_dict=ProxyUserAPIKeyAuth(api_key="sk-test"), + logging_obj=self._logging_obj(), + version=None, + proxy_logging_obj=ProxyLogging(user_api_key_cache=DualCache()), + ) + + assert headers.get("x-litellm-response-cost") == expected_header + + @pytest.mark.asyncio + async def test_an_overflowing_cost_breakdown_sends_no_infinite_cost_header(self): + response: Final = litellm.ModelResponse() + response._hidden_params = {"response_cost": 0.0} + logging_obj: Final = self._logging_obj() + logging_obj.cost_breakdown = { + "original_cost": float("inf"), + "input_cost": float("inf"), + "output_cost": 0.0012, + "total_cost": float("inf"), + } + + headers: Final = await ProxyBaseLLMRequestProcessing.build_litellm_proxy_success_headers_from_llm_response( + response=response, + request_data={"model": "gpt-5.4-mini"}, + request=Request({"type": "http", "method": "POST", "headers": []}), + user_api_key_dict=ProxyUserAPIKeyAuth(api_key="sk-test"), + logging_obj=logging_obj, + version=None, + proxy_logging_obj=ProxyLogging(user_api_key_cache=DualCache()), + ) + + cost_headers: Final = {name: value for name, value in headers.items() if name.startswith("x-litellm-response-cost")} + assert cost_headers == {"x-litellm-response-cost-output": "0.0012"} + def test_a_fresh_zero_cost_v1_messages_call_still_sends_no_cost_header(self): logging_obj: Final = self._logging_obj() logging_obj.model_call_details["response_cost"] = 0.0 diff --git a/tests/unit/test_main.py b/tests/unit/test_main.py index b0bf1cb879e..aaec486b60f 100644 --- a/tests/unit/test_main.py +++ b/tests/unit/test_main.py @@ -5900,6 +5900,28 @@ def test_stream_chunk_builder_keeps_provider_reported_usage_cost(): assert response._hidden_params["response_cost"] == pytest.approx(0.5) +@pytest.mark.parametrize("reported_cost", [float("nan"), float("inf")]) +def test_stream_chunk_builder_bills_the_token_price_when_the_provider_cost_is_not_finite(reported_cost: float): + usage_chunk: Final = _stream_builder_text_chunk("gpt-4o", "") + usage_chunk.usage = Usage(prompt_tokens=10, completion_tokens=5, total_tokens=15, cost=reported_cost) + chunks: Final = [ + _stream_builder_text_chunk("gpt-4o", "Hello "), + _stream_builder_text_chunk("gpt-4o", "world.", finish_reason="stop"), + usage_chunk, + ] + + response: Final = litellm.stream_chunk_builder( + chunks=chunks, messages=[{"role": "user", "content": "hi"}], logging_obj=_stream_builder_logging_obj() + ) + + assert response is not None + prompt_cost, completion_cost = litellm.cost_per_token(model="gpt-4o", usage_object=response.usage) + token_price: Final = prompt_cost + completion_cost + assert token_price > 0 + assert getattr(response.usage, "cost", None) == pytest.approx(token_price) + assert response._hidden_params["response_cost"] == pytest.approx(token_price) + + def test_completion_cost_on_a_stamped_stream_still_bills_the_reported_charge(): usage_chunk: Final = _stream_builder_text_chunk("gpt-4o", "") usage_chunk.usage = Usage(prompt_tokens=10, completion_tokens=5, total_tokens=15, cost=0.5) From 5d5e3685c1d52841082c199a88f8190f35e7be9b Mon Sep 17 00:00:00 2001 From: tessellark <349031000+tessellark@users.noreply.github.com> Date: Fri, 30 Oct 2026 08:48:43 -0700 Subject: [PATCH 07/18] fix(logging): warn when drop_params drops a requested reasoning level on the remaining paths (#48933) * fix(logging): warn when drop_params drops a requested reasoning level on the remaining paths * fix(logging): keep guardrail and token-count reads of a request from logging drop warnings Only the real Anthropic-messages adapter request reports a dropped thinking budget. Guardrail, shadow-eval and token-count translations of the same request now drop it silently, so one request logs one warning. The 422 retry also no longer hashes each loc entry, so a list inside loc cannot raise TypeError. * fix(bedrock): keep the DeepSeek R1 Converse reasoning_effort drop at debug level Only allowed_openai_params reaches that branch, and it puts reasoning_effort back after the provider mapping, so a warning there claimed a drop that never happened * fix(logging): warn on Gemini non-integer budget drops and keep the auto-router baseline silent Gemini 2.x now logs the drop of a non-integer thinking budget. The auto-router savings baseline builds a body it never sends, so it no longer warns. Logged values are capped with reprlib, Azure AI marks only its drop_params retry as gated, and Bedrock Chat Completions names the model without its route prefix. * fix(logging): warn on the Gemini 3, Magistral, Ollama, Fireworks and DeepSeek ARN reasoning drops Gemini 3 and later now log a dropped thinking budget, Magistral logs the level it swaps for a reasoning prompt, Ollama logs a level it maps to think off, Fireworks logs chat_template_kwargs effort keys dropped for a model the map does not list as reasoning, and DeepSeek on Bedrock logs a drop when the model is addressed by an ARN that advertises the param. * fix(logging): warn on the remaining Ollama, DeepSeek and Tencent reasoning drops The ollama/ completion route now shares the chat route's think mapping, so a level it turns off logs the same warning. DeepSeek warns when it drops thinking.budget_tokens, and Tencent warns when an adaptive-only model gets a non-integer budget. Magistral no longer warns on low, medium, high or enabled thinking, the values its reasoning prompt stands in for. The Gemini 3 budget warning no longer claims the model uses its default level. * fix(reasoning): warn when Claude drops a non-string reasoning_effort Claude chat and native /v1/messages dropped a reasoning_effort like 5 or {"effort": 5} with no log line. Both now log one warn-only WARNING. A Magistral thinking.type sent as a list no longer raises TypeError. --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> --- .../reasoning_effort_utils.py | 77 ++++++++++- litellm/llms/anthropic/chat/transformation.py | 2 + litellm/llms/anthropic/common_utils.py | 33 +++-- .../messages/adapters/transformation.py | 18 ++- .../llms/anthropic/messages/native/handler.py | 8 +- .../messages/native/transformation.py | 11 +- .../responses_adapters/transformation.py | 4 +- .../llms/anthropic/prompt_cache_prediction.py | 8 +- litellm/llms/azure_ai/chat/transformation.py | 25 +++- .../chat/chat_completions/transformation.py | 17 +++ .../bedrock/chat/converse_transformation.py | 17 ++- litellm/llms/deepseek/chat/transformation.py | 13 ++ litellm/llms/fal_ai/chat/transformation.py | 13 +- .../llms/fireworks_ai/chat/transformation.py | 14 +- .../fireworks_ai/completion/transformation.py | 7 +- .../llms/hosted_vllm/chat/transformation.py | 2 +- litellm/llms/mistral/chat/transformation.py | 25 +++- litellm/llms/oci/chat/transformation.py | 16 +++ litellm/llms/ollama/chat/transformation.py | 8 +- litellm/llms/ollama/common_utils.py | 12 ++ .../llms/ollama/completion/transformation.py | 8 +- litellm/llms/openai/common_utils.py | 22 +-- litellm/llms/openai/openai.py | 8 +- litellm/llms/tencent/chat/transformation.py | 7 + .../vertex_and_google_ai_studio_gemini.py | 26 +++- .../reasoning_effort_capability.py | 8 +- .../test_anthropic_guardrail_handler.py | 16 +++ .../test_anthropic_chat_transformation.py | 42 ++++++ ...hropic_messages_adapters_transformation.py | 16 ++- ..._handler_reasoning_effort_normalization.py | 26 ++-- .../native/test_anthropic_messages_handler.py | 10 +- .../native/test_anthropic_messages_router.py | 17 ++- .../test_reasoning_effort_translation.py | 46 +++++++ .../test_responses_adapters_transformation.py | 33 +++-- .../messages/test_reasoning_effort_fields.py | 5 +- .../anthropic/test_anthropic_common_utils.py | 18 ++- .../test_anthropic_prompt_cache_prediction.py | 40 ++++++ .../chat/test_azure_ai_transformation.py | 61 +++++++++ ...bedrock_chat_completions_transformation.py | 52 ++++--- .../chat/test_converse_transformation.py | 47 +++++++ .../chat/test_deepseek_chat_transformation.py | 41 ++++++ .../chat/test_fal_ai_chat_transformation.py | 33 ++++- .../test_fireworks_ai_chat_transformation.py | 35 +++++ ...works_ai_text_completion_transformation.py | 37 +++++ .../test_mistral_chat_transformation.py | 105 +++++++++++++++ .../oci/chat/test_oci_chat_transformation.py | 41 +++--- .../ollama/test_ollama_chat_transformation.py | 35 +++++ .../test_ollama_completion_transformation.py | 39 ++++++ .../llms/openai/test_openai_common_utils.py | 127 +++++++++++++++++- .../chat/test_tencent_chat_transformation.py | 26 ++++ ...test_vertex_and_google_ai_studio_gemini.py | 72 ++++++++++ tests/unit/test_main.py | 36 ++++- 52 files changed, 1318 insertions(+), 147 deletions(-) diff --git a/litellm/litellm_core_utils/reasoning_effort_utils.py b/litellm/litellm_core_utils/reasoning_effort_utils.py index acd20aeb37d..e9c94e32f8f 100644 --- a/litellm/litellm_core_utils/reasoning_effort_utils.py +++ b/litellm/litellm_core_utils/reasoning_effort_utils.py @@ -1,6 +1,12 @@ +import reprlib +from collections.abc import Collection, Mapping +from dataclasses import dataclass from typing import Final, Literal +from pydantic import TypeAdapter + import litellm +from litellm._logging import verbose_logger from litellm.constants import ( DEFAULT_REASONING_EFFORT_HIGH_THINKING_BUDGET, DEFAULT_REASONING_EFFORT_LOW_THINKING_BUDGET, @@ -11,6 +17,54 @@ from litellm.litellm_core_utils.core_helpers import as_str_mapping OpenAIStyleReasoningEffort = Literal["minimal", "low", "medium", "high"] +_DROP_WARNING: Final = "Dropping %s=%s for model=%r because %s" +_DROP_PARAMS_DROP_WARNING: Final = "Dropping %s=%s for model=%r (drop_params=True) because %s" +_NON_NUMERIC_BUDGET_REASON: Final = "it is not a number. Send budget_tokens as an integer to set a reasoning level." +_NON_STRING_EFFORT_REASON: Final = ( + "it is not a string. Send reasoning_effort as a level name such as 'high' to set a reasoning level." +) +_REASONING_PARAMS: Final = ("reasoning", "reasoning_effort", "thinking") +OPAQUE_VALUE: Final = TypeAdapter(object) + + +def warn_dropped_reasoning( + param: str, value: object, model: str | None, reason: str, *, drop_params_gated: bool +) -> None: + verbose_logger.warning( + _DROP_PARAMS_DROP_WARNING if drop_params_gated else _DROP_WARNING, param, reprlib.repr(value), model, reason + ) + + +def warn_non_string_reasoning_effort(value: object, model: str | None) -> None: + effort: Final = as_str_mapping(value) + requested_level: Final = effort.get("effort") if effort is not None else value + if requested_level is None or isinstance(value, str): + return + warn_dropped_reasoning("reasoning_effort", value, model, _NON_STRING_EFFORT_REASON, drop_params_gated=False) + + +def warn_dropped_reasoning_params( + dropped: Collection[object], + request: Mapping[str, object], + model: str | None, + reason: str, + *, + drop_params_gated: bool, +) -> None: + for param in _REASONING_PARAMS: + if param in dropped and param in request: + warn_dropped_reasoning(param, request[param], model, reason, drop_params_gated=drop_params_gated) + + +@dataclass(frozen=True, slots=True) +class DroppedThinkingBudget: + budget_tokens: object + + def warn(self, model: str) -> None: + warn_dropped_reasoning( + "thinking.budget_tokens", self.budget_tokens, model, _NON_NUMERIC_BUDGET_REASON, drop_params_gated=True + ) + def reasoning_effort_from_thinking_budget( budget_tokens: float, @@ -29,7 +83,9 @@ def reasoning_effort_from_thinking_budget( return "minimal" -def reasoning_effort_from_thinking(thinking: object, *, drop_params: bool) -> OpenAIStyleReasoningEffort | None: +def _effort_or_dropped_budget( + thinking: object, *, drop_params: bool +) -> OpenAIStyleReasoningEffort | DroppedThinkingBudget | None: fields: Final = as_str_mapping(thinking) if fields is None or fields.get("type") != "enabled": return None @@ -37,10 +93,27 @@ def reasoning_effort_from_thinking(thinking: object, *, drop_params: bool) -> Op if isinstance(budget, (int, float)) and not isinstance(budget, bool): return reasoning_effort_from_thinking_budget(budget) if drop_params or litellm.drop_params is True: - return None + return DroppedThinkingBudget(budget) raise non_numeric_thinking_budget_error(budget) +def reasoning_effort_from_thinking( + thinking: object, *, model: str, drop_params: bool +) -> OpenAIStyleReasoningEffort | None: + effort: Final = _effort_or_dropped_budget(thinking, drop_params=drop_params) + if not isinstance(effort, DroppedThinkingBudget): + return effort + effort.warn(model) + return None + + +def reasoning_effort_from_thinking_silently( + thinking: object, *, drop_params: bool +) -> OpenAIStyleReasoningEffort | None: + effort: Final = _effort_or_dropped_budget(thinking, drop_params=drop_params) + return None if isinstance(effort, DroppedThinkingBudget) else effort + + def non_numeric_thinking_budget_error(budget: object) -> UnsupportedParamsError: return UnsupportedParamsError( message=( diff --git a/litellm/llms/anthropic/chat/transformation.py b/litellm/llms/anthropic/chat/transformation.py index 865247afc63..d773e31c733 100644 --- a/litellm/llms/anthropic/chat/transformation.py +++ b/litellm/llms/anthropic/chat/transformation.py @@ -48,6 +48,7 @@ from litellm.litellm_core_utils.prompt_templates.mid_conversation_system import place_mid_conversation_system, split_leading_system_run, ) +from litellm.litellm_core_utils.reasoning_effort_utils import OPAQUE_VALUE, warn_non_string_reasoning_effort from litellm.llms.base_llm.base_utils import ( ResponseFormatSchema, as_json_object, @@ -1951,6 +1952,7 @@ class AnthropicConfig(AnthropicModelInfo, BaseConfig): if isinstance(effort_value, dict): effort_value = effort_value.get("effort") if not isinstance(effort_value, str): + warn_non_string_reasoning_effort(OPAQUE_VALUE.validate_python(value), model) continue mapped_thinking = AnthropicConfig._map_reasoning_effort( reasoning_effort=effort_value, diff --git a/litellm/llms/anthropic/common_utils.py b/litellm/llms/anthropic/common_utils.py index 82977f97854..6d1ac24af55 100644 --- a/litellm/llms/anthropic/common_utils.py +++ b/litellm/llms/anthropic/common_utils.py @@ -37,7 +37,7 @@ from litellm.litellm_core_utils.prompt_templates.mid_conversation_system import message_field, parts_of, ) -from litellm.litellm_core_utils.reasoning_effort_utils import non_numeric_thinking_budget_error +from litellm.litellm_core_utils.reasoning_effort_utils import DroppedThinkingBudget, non_numeric_thinking_budget_error from litellm.litellm_core_utils.url_utils import join_api_base from litellm.llms.anthropic.wif import ( aget_anthropic_wif_token, @@ -610,11 +610,11 @@ def _numeric_budget_tokens(budget_tokens: object) -> int | None: return None -def _legacy_budget_tokens(budget_tokens: object, *, drop_params: bool) -> int | None: +def _legacy_budget_tokens(budget_tokens: object, *, drop_params: bool) -> int | DroppedThinkingBudget: if (numeric := _numeric_budget_tokens(budget_tokens)) is not None: return numeric if drop_params or litellm.drop_params is True: - return None + return DroppedThinkingBudget(budget_tokens) raise non_numeric_thinking_budget_error(budget_tokens) @@ -1097,6 +1097,22 @@ class AnthropicModelInfo(BaseLLMModelInfo): custom_llm_provider: str, drop_params: bool, ) -> None: + dropped: Final = AnthropicModelInfo.translate_legacy_thinking_for_adaptive_model_silently( + model=model, + optional_params=optional_params, + custom_llm_provider=custom_llm_provider, + drop_params=drop_params, + ) + if dropped is not None: + dropped.warn(model) + + @staticmethod + def translate_legacy_thinking_for_adaptive_model_silently( + model: str, + optional_params: MutableMapping[str, object], # mutable-ok: in-place out-param like the sibling helpers + custom_llm_provider: str, + drop_params: bool, + ) -> DroppedThinkingBudget | None: """Translate legacy ``thinking.type=enabled`` to adaptive for the adaptive-thinking models that reject it (4.7+ and the 5 families). Models flagged ``supports_legacy_thinking`` (the 4.6 family) accept the @@ -1105,12 +1121,12 @@ class AnthropicModelInfo(BaseLLMModelInfo): ``output_config.effort`` is never overridden. """ if not AnthropicModelInfo._is_adaptive_thinking_model(model, custom_llm_provider): - return + return None if AnthropicModelInfo._supports_legacy_thinking(model, custom_llm_provider): - return + return None thinking: Final = validated_object_mapping(optional_params.get("thinking")) if thinking is None or thinking.get("type") != "enabled": - return + return None budget_tokens: Final = _legacy_budget_tokens(thinking.get("budget_tokens", 0), drop_params=drop_params) display: Final = thinking.get("display") @@ -1119,8 +1135,8 @@ class AnthropicModelInfo(BaseLLMModelInfo): thinking, rebuilt_keys=frozenset(("type", "budget_tokens", "display")), drop_params=drop_params ) optional_params["thinking"] = {**carried_keys, "type": "adaptive", **supported_display} - if budget_tokens is None: - return + if isinstance(budget_tokens, DroppedThinkingBudget): + return budget_tokens effort: Final = AnthropicModelInfo._legacy_budget_to_effort( model=model, budget_tokens=budget_tokens, custom_llm_provider=custom_llm_provider ) @@ -1129,6 +1145,7 @@ class AnthropicModelInfo(BaseLLMModelInfo): "effort": effort, **(existing_output_config if isinstance(existing_output_config, dict) else MappingProxyType({})), } + return None @staticmethod def _legacy_budget_to_effort(model: str, budget_tokens: int, custom_llm_provider: str) -> str: diff --git a/litellm/llms/anthropic/messages/adapters/transformation.py b/litellm/llms/anthropic/messages/adapters/transformation.py index 8ea6293bbe6..bc9649ed867 100644 --- a/litellm/llms/anthropic/messages/adapters/transformation.py +++ b/litellm/llms/anthropic/messages/adapters/transformation.py @@ -134,6 +134,7 @@ from litellm.litellm_core_utils.prompt_templates.factory import ( ) from litellm.litellm_core_utils.reasoning_effort_utils import ( reasoning_effort_from_thinking, + reasoning_effort_from_thinking_silently, ) from litellm.llms.anthropic.common_utils import ( eager_input_streaming_flag, @@ -297,6 +298,7 @@ class AnthropicAdapter: custom_llm_provider=custom_llm_provider, api_base=api_base, drop_params=drop_params, + report_reasoning_drops=True, ) return translated_body, tool_name_mapping @@ -666,7 +668,9 @@ class LiteLLMAnthropicMessagesAdapter: def translate_anthropic_thinking_to_reasoning_effort( thinking: AnthropicThinkingParam, *, + model: str, drop_params: bool = True, + report_reasoning_drops: bool = False, ) -> str | None: """ Translate Anthropic's thinking parameter to OpenAI's reasoning_effort. @@ -685,7 +689,9 @@ class LiteLLMAnthropicMessagesAdapter: if thinking_type == "disabled": return "none" elif thinking_type == "enabled": - return reasoning_effort_from_thinking(thinking, drop_params=drop_params) + if report_reasoning_drops: + return reasoning_effort_from_thinking(thinking, model=model, drop_params=drop_params) + return reasoning_effort_from_thinking_silently(thinking, drop_params=drop_params) elif thinking_type == "adaptive": # Adaptive thinking: effort is controlled by output_config.effort, # not budget_tokens. Return a default; caller should override with @@ -752,7 +758,7 @@ class LiteLLMAnthropicMessagesAdapter: return {"thinking": thinking} else: reasoning_effort: Final = LiteLLMAnthropicMessagesAdapter.translate_anthropic_thinking_to_reasoning_effort( - thinking + thinking, model=model ) if reasoning_effort: return { @@ -1147,6 +1153,7 @@ class LiteLLMAnthropicMessagesAdapter: *, custom_llm_provider: str | None = None, drop_params: bool = True, + report_reasoning_drops: bool = False, ) -> None: """Translate Anthropic thinking to either thinking or reasoning_effort. @@ -1207,7 +1214,10 @@ class LiteLLMAnthropicMessagesAdapter: return reasoning_effort: Final = declared_effort or self.translate_anthropic_thinking_to_reasoning_effort( - cast(AnthropicThinkingParam, thinking), drop_params=drop_params + cast(AnthropicThinkingParam, thinking), + model=model, + drop_params=drop_params, + report_reasoning_drops=report_reasoning_drops, ) if not reasoning_effort: return @@ -1267,6 +1277,7 @@ class LiteLLMAnthropicMessagesAdapter: api_base: str | None = None, preserve_midturn_system: bool = False, drop_params: bool = True, + report_reasoning_drops: bool = False, ) -> tuple[ChatCompletionRequest, dict[str, str]]: """ This is used by the beta Anthropic Adapter, for translating anthropic `/v1/messages` requests to the openai format. @@ -1321,6 +1332,7 @@ class LiteLLMAnthropicMessagesAdapter: new_kwargs=new_kwargs, custom_llm_provider=custom_llm_provider, drop_params=drop_params, + report_reasoning_drops=report_reasoning_drops, ) ## CONVERT STOP_SEQUENCES self._translate_stop_sequences_to_openai( diff --git a/litellm/llms/anthropic/messages/native/handler.py b/litellm/llms/anthropic/messages/native/handler.py index 6f97a019a96..1da7c2efdda 100644 --- a/litellm/llms/anthropic/messages/native/handler.py +++ b/litellm/llms/anthropic/messages/native/handler.py @@ -20,6 +20,7 @@ from litellm.litellm_core_utils.core_helpers import as_str_mapping from litellm.litellm_core_utils.exception_mapping_utils import exception_type from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj from litellm.llms.anthropic.common_utils import ( + AnthropicModelInfo, strip_provider_specific_fields_from_anthropic_messages, tools_marked_for_a_tool_config_point, ) @@ -36,7 +37,6 @@ from litellm.router_utils.reasoning_effort_capability import ( keeps_effort, model_group_from_metadata, resolve_deployment_reasoning_effort, - with_adaptive_thinking_as_effort, ) from litellm.types.llms.anthropic import AllAnthropicToolsValues from litellm.types.llms.anthropic_messages.anthropic_request import AnthropicMetadata @@ -185,9 +185,11 @@ def _with_deployment_output_effort( ).value def translate(source: Mapping[str, object]) -> Mapping[str, object]: - return with_adaptive_thinking_as_effort( - source, model=model, custom_llm_provider=custom_llm_provider, drop_params=drop_params + translated: Final = dict(source) + AnthropicModelInfo.translate_legacy_thinking_for_adaptive_model( + model=model, optional_params=translated, custom_llm_provider=custom_llm_provider, drop_params=drop_params ) + return translated requested_output: Final = as_str_mapping(params.get("output_config")) caller_output: Final = None if requested_output is None else resolve(requested_output) diff --git a/litellm/llms/anthropic/messages/native/transformation.py b/litellm/llms/anthropic/messages/native/transformation.py index ef855eff180..c148a0defe8 100644 --- a/litellm/llms/anthropic/messages/native/transformation.py +++ b/litellm/llms/anthropic/messages/native/transformation.py @@ -21,6 +21,7 @@ from litellm.constants import ANTHROPIC_BILLING_METADATA_PREFIX from litellm.exceptions import AuthenticationError from litellm.litellm_core_utils.core_helpers import drops_unsupported_params from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj +from litellm.litellm_core_utils.reasoning_effort_utils import OPAQUE_VALUE, warn_non_string_reasoning_effort from litellm.litellm_core_utils.url_utils import ensure_api_route from litellm.llms.anthropic.messages.convertible_blocks import drop_non_string_tool_result_texts from litellm.llms.anthropic.messages.safeguards import proxy_serves_safeguards @@ -103,6 +104,7 @@ _MESSAGES_ADAPTER: Final = TypeAdapter(list[dict[str, object]]) class AnthropicMessagesConfig(BaseAnthropicMessagesConfig): _workload_identity_eligible: ClassVar[bool] = True + _report_reasoning_drops: ClassVar[bool] = True @property def custom_llm_provider(self) -> str | None: @@ -695,12 +697,17 @@ class AnthropicMessagesConfig(BaseAnthropicMessagesConfig): ): anthropic_messages_optional_request_params.pop("output_format") # pyright: ignore[reportUnknownMemberType] # the request params arrive as a bare dict + requested_reasoning_effort: Final = OPAQUE_VALUE.validate_python( + anthropic_messages_optional_request_params.get("reasoning_effort") # pyright: ignore[reportUnknownMemberType] # the request params arrive as a bare dict + ) self._translate_reasoning_effort_to_anthropic( model=model, optional_params=anthropic_messages_optional_request_params, max_tokens=max_tokens, custom_llm_provider=self._resolved_provider, ) + if self._report_reasoning_drops: + warn_non_string_reasoning_effort(requested_reasoning_effort, model) AnthropicModelInfo.maybe_drop_disabled_thinking( model=model, @@ -708,12 +715,14 @@ class AnthropicMessagesConfig(BaseAnthropicMessagesConfig): custom_llm_provider=self._resolved_provider, ) - AnthropicModelInfo.translate_legacy_thinking_for_adaptive_model( + dropped_budget: Final = AnthropicModelInfo.translate_legacy_thinking_for_adaptive_model_silently( model=model, optional_params=anthropic_messages_optional_request_params, custom_llm_provider=self._resolved_provider, drop_params=getattr(litellm_params, "drop_params", None) is True, ) + if dropped_budget is not None and self._report_reasoning_drops: + dropped_budget.warn(model) self._translate_adaptive_effort_for_non_adaptive_model( model=model, diff --git a/litellm/llms/anthropic/messages/responses_adapters/transformation.py b/litellm/llms/anthropic/messages/responses_adapters/transformation.py index 94c608f90e6..0630bef0311 100644 --- a/litellm/llms/anthropic/messages/responses_adapters/transformation.py +++ b/litellm/llms/anthropic/messages/responses_adapters/transformation.py @@ -534,6 +534,7 @@ class LiteLLMAnthropicToResponsesAPIAdapter: output_config: dict[str, object] | None = None, auto_summary_level: str = "detailed", *, + model: str, drop_params: bool = True, ) -> dict[str, object] | None: """ @@ -555,7 +556,7 @@ class LiteLLMAnthropicToResponsesAPIAdapter: if isinstance(output_config, dict) and output_config.get("effort"): effort = output_config["effort"] elif thinking_type == "enabled": - budget_effort: Final = reasoning_effort_from_thinking(thinking, drop_params=drop_params) + budget_effort: Final = reasoning_effort_from_thinking(thinking, model=model, drop_params=drop_params) if budget_effort is None: return None effort = budget_effort @@ -667,6 +668,7 @@ class LiteLLMAnthropicToResponsesAPIAdapter: thinking, output_config=cast(dict[str, object] | None, output_config), auto_summary_level=auto_summary_level, + model=model, drop_params=drop_params, ) if reasoning: diff --git a/litellm/llms/anthropic/prompt_cache_prediction.py b/litellm/llms/anthropic/prompt_cache_prediction.py index 27e003ce8b6..2da071ebdd2 100644 --- a/litellm/llms/anthropic/prompt_cache_prediction.py +++ b/litellm/llms/anthropic/prompt_cache_prediction.py @@ -9,7 +9,7 @@ from dataclasses import dataclass, field from functools import reduce from itertools import accumulate, groupby from types import MappingProxyType -from typing import Annotated, Final, Literal, Protocol, TypeAlias +from typing import Annotated, ClassVar, Final, Literal, Protocol, TypeAlias import httpx from pydantic import ConfigDict, Field, JsonValue, StrictInt, TypeAdapter, ValidationError @@ -81,6 +81,10 @@ _DEPLOYMENT_OPTIONS: Final = frozenset( ) +class _BaselineAnthropicMessagesConfig(AnthropicMessagesConfig): + _report_reasoning_drops: ClassVar[bool] = False + + class _StrictModel(LiteLLMBaseModel): model_config = ConfigDict(extra="forbid", frozen=True, strict=True) @@ -661,7 +665,7 @@ def prepare_native_baseline_body(request: Mapping[str, object], model: str) -> M TypeAdapter(tuple[str, ...]).validate_python(owned.get("additional_drop_params") or ()), dict(options), ) - config: Final = AnthropicMessagesConfig() + config: Final = _BaselineAnthropicMessagesConfig() body: Final = config.transform_anthropic_messages_request( model=resolved_model, messages=strip_provider_specific_fields_from_anthropic_messages(messages), diff --git a/litellm/llms/azure_ai/chat/transformation.py b/litellm/llms/azure_ai/chat/transformation.py index 5381447f28f..5214d31fb2e 100644 --- a/litellm/llms/azure_ai/chat/transformation.py +++ b/litellm/llms/azure_ai/chat/transformation.py @@ -20,6 +20,7 @@ from litellm.litellm_core_utils.prompt_templates.common_utils import ( prompt_cache_breakpoint_for_wire, with_prompt_cache_breakpoint, ) +from litellm.litellm_core_utils.reasoning_effort_utils import warn_dropped_reasoning_params from litellm.llms.azure.common_utils import BaseAzureLLM from litellm.llms.azure_ai.common_utils import ( api_key_header_for_base, @@ -40,6 +41,14 @@ if TYPE_CHECKING: from litellm.litellm_core_utils.tokenizer import Encoding as Tokenizer +_REQUEST_BODY: Final = TypeAdapter(dict[str, object]) + + +def _request_model(request: Mapping[str, object]) -> str | None: + model: Final = request.get("model") + return model if isinstance(model, str) else None + + class AzureFoundryErrorStrings(str, enum.Enum): SET_EXTRA_PARAMETERS_TO_PASS_THROUGH = "Set extra-parameters to 'pass-through'" @@ -510,7 +519,13 @@ class AzureAIStudioConfig(OpenAIConfig): request_data = self._drop_extra_params_from_request_data(request_data, error_text) if "Extra inputs are not permitted" in error_text and self._error_has_tool_level_extra_fields(error_text): request_data = self._drop_tool_level_extra_fields(request_data, error_text) - data: Final = drop_params_from_unprocessable_entity_error(e=e, data=request_data) + data: Final = drop_params_from_unprocessable_entity_error( + e=e, + data=request_data, + model=_request_model(_REQUEST_BODY.validate_python(request_data)), + drop_params_gated="Extra inputs are not permitted" in error_text + and not self._error_has_tool_level_extra_fields(error_text), + ) return data def _drop_tool_level_extra_fields(self, request_data: dict, error_text: str) -> dict: @@ -526,6 +541,14 @@ class AzureAIStudioConfig(OpenAIConfig): def _drop_extra_params_from_request_data(self, request_data: dict, error_text: str) -> dict: params_to_drop: Final = self._extract_params_to_drop_from_error_text(error_text) if params_to_drop: + request: Final = _REQUEST_BODY.validate_python(request_data) + warn_dropped_reasoning_params( + params_to_drop, + request, + _request_model(request), + "Azure AI rejected it as an extra parameter its deployment does not pass through.", + drop_params_gated=False, + ) for param in params_to_drop: if param in request_data: request_data.pop(param, None) diff --git a/litellm/llms/bedrock/chat/chat_completions/transformation.py b/litellm/llms/bedrock/chat/chat_completions/transformation.py index 123e4a9e6b4..dd2277a6c35 100644 --- a/litellm/llms/bedrock/chat/chat_completions/transformation.py +++ b/litellm/llms/bedrock/chat/chat_completions/transformation.py @@ -33,6 +33,7 @@ from litellm.litellm_core_utils.prompt_templates.image_handling import ( inline_remote_image_urls, inline_remote_media, ) +from litellm.litellm_core_utils.reasoning_effort_utils import warn_dropped_reasoning from litellm.litellm_core_utils.url_utils import ensure_api_route, join_api_base from litellm.llms.base_llm.chat.transformation import BaseLLMException from litellm.llms.bedrock.base_aws_llm import BaseAWSLLM, bedrock_bearer_token @@ -42,6 +43,7 @@ from litellm.llms.bedrock.common_utils import ( bedrock_rejects_stop_sequences, bedrock_runtime_chat_completions_serves_reasoning_inline, split_bedrock_region_path, + strip_bedrock_routing_prefix, ) from litellm.llms.openai.chat.gpt_transformation import OpenAIChatCompletionStreamingHandler from litellm.llms.openai_like.chat.transformation import OpenAILikeChatConfig @@ -121,6 +123,13 @@ def without_refused_reasoning_effort(model: str, params: Mapping[str, object]) - effort: Final = params.get("reasoning_effort") if not isinstance(effort, str) or effort not in chat_completions_reasoning_efforts_refused_for(model): return params + warn_dropped_reasoning( + "reasoning_effort", + effort, + strip_bedrock_routing_prefix(model), + "AWS's Chat Completions endpoint rejects that level for this model family, so AWS applies its default effort.", + drop_params_gated=False, + ) return _without_params(params, frozenset(("reasoning_effort",))) @@ -445,6 +454,14 @@ class AmazonBedrockRuntimeChatCompletionsConfig(OpenAILikeChatConfig): ), status_code=400, ) + if malformed_effort: + warn_dropped_reasoning( + "reasoning_effort", + raw_params["reasoning_effort"], + strip_bedrock_routing_prefix(model), + "Bedrock's Chat Completions endpoint takes it only as a string, so AWS applies its default effort.", + drop_params_gated=True, + ) rejected_stop: Final = frozenset(("stop",)) if bedrock_rejects_stop_sequences(model) else frozenset[str]() return dict( without_refused_reasoning_effort( diff --git a/litellm/llms/bedrock/chat/converse_transformation.py b/litellm/llms/bedrock/chat/converse_transformation.py index 5e621a5f322..3800489ec05 100644 --- a/litellm/llms/bedrock/chat/converse_transformation.py +++ b/litellm/llms/bedrock/chat/converse_transformation.py @@ -50,6 +50,7 @@ from litellm.litellm_core_utils.prompt_templates.mid_conversation_system import message_field, parts_of, ) +from litellm.litellm_core_utils.reasoning_effort_utils import OPAQUE_VALUE, warn_dropped_reasoning from litellm.llms.anthropic.chat.transformation import ( DROP_UNSUPPORTED_ADAPTIVE_THINKING_WARNING, DROP_UNSUPPORTED_OUTPUT_CONFIG_WARNING, @@ -187,6 +188,8 @@ def _tools_the_model_accepts( _OUTPUT_CONFIG_BLOCK: Final = TypeAdapter(OutputConfigBlock) +_DEEPSEEK_THINKING_REASON: Final = "DeepSeek on Bedrock reasons natively and takes no thinking field." +_DEEPSEEK_R1_EFFORT_REASON: Final = "DeepSeek R1 on Bedrock always reasons and rejects reasoning_effort." def _has_native_text_format(output_config: object) -> bool: @@ -640,6 +643,10 @@ class AmazonConverseConfig(BaseConfig): or supports_reasoning(model=base_model, custom_llm_provider=self.custom_llm_provider) ) + def _warn_native_reasoning_drop(self, param: str, value: object, model: str, reason: str) -> None: + if param in self.get_supported_openai_params(model): + warn_dropped_reasoning(param, value, model, reason, drop_params_gated=False) + def _model_rejects_reasoning_effort_param(self, model: str, base_model: str) -> bool: """Whether the model returns a 400 for every ``reasoning_effort`` shape on Converse. @@ -1143,9 +1150,8 @@ class AmazonConverseConfig(BaseConfig): "tool_choice": {"type": "auto", "disable_parallel_tool_use": not value} } if param == "thinking" and drop_thinking_param: - verbose_logger.debug( - "Dropping unsupported `thinking` param for Bedrock model=%s; it reasons natively.", - model, + self._warn_native_reasoning_drop( + param, OPAQUE_VALUE.validate_python(value), model, _DEEPSEEK_THINKING_REASON ) elif param == "thinking" and not self._is_openai_gpt_reasoning_model(model): if ( @@ -1179,10 +1185,7 @@ class AmazonConverseConfig(BaseConfig): drop_params=drop_params, ) elif param == "reasoning_effort" and isinstance(value, str) and drop_reasoning_effort_param: - verbose_logger.debug( - "Dropping unsupported `reasoning_effort` param for Bedrock model=%s; it always reasons and rejects it.", - model, - ) + self._warn_native_reasoning_drop(param, value, model, _DEEPSEEK_R1_EFFORT_REASON) elif ( param == "reasoning_effort" and isinstance(value, str) diff --git a/litellm/llms/deepseek/chat/transformation.py b/litellm/llms/deepseek/chat/transformation.py index ecaa0a2ef16..5b7c2a2c410 100644 --- a/litellm/llms/deepseek/chat/transformation.py +++ b/litellm/llms/deepseek/chat/transformation.py @@ -6,10 +6,12 @@ from collections.abc import Coroutine, Mapping, Sequence from typing import Final, Literal, cast, overload import litellm +from litellm.litellm_core_utils.core_helpers import as_str_mapping from litellm.litellm_core_utils.prompt_templates.common_utils import ( convert_content_list_to_str, extract_search_results_text, ) +from litellm.litellm_core_utils.reasoning_effort_utils import OPAQUE_VALUE, warn_dropped_reasoning from litellm.litellm_core_utils.url_utils import ensure_api_route from litellm.secret_managers.main import get_secret_str from litellm.types.llms.openai import AllMessageValues @@ -18,6 +20,16 @@ from litellm.utils import supports_reasoning, supports_vision from ...openai.chat.gpt_transformation import OpenAIGPTConfig SUPPORTED_REASONING_EFFORTS: Final = ("minimal", "low", "medium", "high", "xhigh", "max", "ultra") +_NO_BUDGET_REASON: Final = "DeepSeek takes thinking only as enabled or disabled, with no budget." + + +def _warn_on_dropped_budget(thinking: object, model: str) -> None: + fields: Final = as_str_mapping(thinking) + if fields is None or fields.get("type") != "enabled" or "budget_tokens" not in fields: + return + warn_dropped_reasoning( + "thinking.budget_tokens", fields["budget_tokens"], model, _NO_BUDGET_REASON, drop_params_gated=False + ) class DeepSeekChatConfig(OpenAIGPTConfig): @@ -53,6 +65,7 @@ class DeepSeekChatConfig(OpenAIGPTConfig): # Then re-add only if valid for DeepSeek thinking_value: Final = optional_params.pop("thinking", None) reasoning_effort: Final = optional_params.pop("reasoning_effort", None) + _warn_on_dropped_budget(OPAQUE_VALUE.validate_python(thinking_value), model) # Handle thinking parameter - accept both enabled and disabled, ignore budget_tokens if isinstance(thinking_value, dict) and thinking_value.get("type") in ("enabled", "disabled"): diff --git a/litellm/llms/fal_ai/chat/transformation.py b/litellm/llms/fal_ai/chat/transformation.py index 115b1f840de..848ae498cdd 100644 --- a/litellm/llms/fal_ai/chat/transformation.py +++ b/litellm/llms/fal_ai/chat/transformation.py @@ -12,7 +12,9 @@ from typing import TYPE_CHECKING, Final import httpx from pydantic import ConfigDict, TypeAdapter +import litellm from litellm.litellm_core_utils.core_helpers import map_finish_reason +from litellm.litellm_core_utils.reasoning_effort_utils import warn_dropped_reasoning from litellm.llms.base_llm.chat.transformation import BaseConfig, BaseLLMException from litellm.secret_managers.main import get_secret_str from litellm.types.llms.base import LiteLLMBaseModel @@ -29,6 +31,10 @@ PROVIDER_PREFIX: Final[str] = "fal_ai/" PASSTHROUGH_PARAMS: Final[frozenset[str]] = frozenset(("reasoning", "temperature", "top_p")) REASONING_DISABLED_EFFORTS: Final[frozenset[str]] = frozenset(("none", "minimal")) REASONING_ENABLED_EFFORTS: Final[frozenset[str]] = frozenset(("low", "medium", "high")) +_UNSUPPORTED_EFFORT_REASON: Final = ( + f"fal only turns reasoning off for {sorted(REASONING_DISABLED_EFFORTS)} " + f"and on for {sorted(REASONING_ENABLED_EFFORTS)}." +) class _FalUsage(LiteLLMBaseModel): @@ -120,9 +126,10 @@ class FalAIChatConfig(BaseConfig): return False if isinstance(value, str) and value in REASONING_ENABLED_EFFORTS: return True - if drop_params: - return None - raise FalAIError(status_code=400, message=f"Unsupported reasoning_effort {value!r} for {model}") + if not (drop_params or litellm.drop_params is True): + raise FalAIError(status_code=400, message=f"Unsupported reasoning_effort {value!r} for {model}") + warn_dropped_reasoning("reasoning_effort", value, model, _UNSUPPORTED_EFFORT_REASON, drop_params_gated=True) + return None def _translate_param(self, param: str, value: object, model: str, drop_params: bool) -> tuple[str, object] | None: if param in ("temperature", "top_p"): diff --git a/litellm/llms/fireworks_ai/chat/transformation.py b/litellm/llms/fireworks_ai/chat/transformation.py index fa5794d71b3..92c6c8439b3 100644 --- a/litellm/llms/fireworks_ai/chat/transformation.py +++ b/litellm/llms/fireworks_ai/chat/transformation.py @@ -17,6 +17,7 @@ from litellm.litellm_core_utils.prompt_templates.common_utils import ( strip_litellm_response_only_fields, unpack_legacy_defs, ) +from litellm.litellm_core_utils.reasoning_effort_utils import OPAQUE_VALUE, warn_dropped_reasoning from litellm.litellm_core_utils.url_utils import join_api_base from litellm.secret_managers.main import get_secret_str from litellm.types.llms.openai import ( @@ -97,6 +98,7 @@ def _json_schema_response_format(schema: object, name: str) -> Mapping[str, obje EFFORT_KWARG_KEYS: Final = frozenset({"enable_thinking", "thinking", "reasoning_budget", "low_effort"}) +_NOT_A_REASONING_MODEL_REASON: Final = "litellm's model map does not list it as a reasoning model." def _bool_from_kwargs(kwargs: Mapping[str, object], keys: tuple[str, ...]) -> bool | None: @@ -107,6 +109,13 @@ def _bool_from_kwargs(kwargs: Mapping[str, object], keys: tuple[str, ...]) -> bo return None +def warn_dropped_effort_kwargs(chat_template_kwargs: object, effort: object, model: str) -> None: + if effort != "none": + warn_dropped_reasoning( + "chat_template_kwargs", chat_template_kwargs, model, _NOT_A_REASONING_MODEL_REASON, drop_params_gated=False + ) + + def effort_from_chat_template_kwargs(kwargs: Mapping[str, object]) -> object: enable_thinking: Final = _bool_from_kwargs(kwargs, ("enable_thinking", "thinking")) if enable_thinking is False: @@ -438,10 +447,7 @@ class FireworksAIConfig(FireworksAIMixin, OpenAIGPTConfig): if effort is None: return () if not supports_reasoning(model=model, custom_llm_provider="fireworks_ai"): - verbose_logger.debug( - "fireworks_ai model %r does not support reasoning; dropping chat_template_kwargs effort keys.", - model, - ) + warn_dropped_effort_kwargs(OPAQUE_VALUE.validate_python(chat_template_kwargs), effort, model) return () return (("reasoning_effort", effort),) diff --git a/litellm/llms/fireworks_ai/completion/transformation.py b/litellm/llms/fireworks_ai/completion/transformation.py index 1f5d4f306a9..611e60cedd4 100644 --- a/litellm/llms/fireworks_ai/completion/transformation.py +++ b/litellm/llms/fireworks_ai/completion/transformation.py @@ -2,6 +2,7 @@ from collections.abc import Mapping from typing import Final from litellm._logging import verbose_logger +from litellm.litellm_core_utils.reasoning_effort_utils import OPAQUE_VALUE from litellm.types.llms.openai import AllMessageValues, OpenAITextCompletionUserMessage from litellm.utils import supports_reasoning @@ -12,6 +13,7 @@ from ..chat.transformation import ( NIM_VLLM_STRIP_PARAMS, FireworksAIConfig, effort_from_chat_template_kwargs, + warn_dropped_effort_kwargs, ) from ..common_utils import FireworksAIMixin, resolve_fireworks_resource_name @@ -129,10 +131,7 @@ class FireworksAITextCompletionConfig(FireworksAIMixin, BaseTextCompletionConfig ) return result if not supports_reasoning(model=model, custom_llm_provider="fireworks_ai"): - verbose_logger.debug( - "fireworks_ai model %r does not support reasoning; dropping chat_template_kwargs effort keys.", - model, - ) + warn_dropped_effort_kwargs(OPAQUE_VALUE.validate_python(chat_template_kwargs), effort, model) return result return {**result, "reasoning_effort": effort} diff --git a/litellm/llms/hosted_vllm/chat/transformation.py b/litellm/llms/hosted_vllm/chat/transformation.py index 63e6cb2b5da..9f9503f93b6 100644 --- a/litellm/llms/hosted_vllm/chat/transformation.py +++ b/litellm/llms/hosted_vllm/chat/transformation.py @@ -99,7 +99,7 @@ class HostedVLLMChatConfig(OpenAIGPTConfig): thinking_effort: Final = ( None if "reasoning_effort" in non_default_params - else reasoning_effort_from_thinking(thinking, drop_params=drop_params) + else reasoning_effort_from_thinking(thinking, model=model, drop_params=drop_params) ) if thinking_effort is not None: non_default_params["reasoning_effort"] = thinking_effort diff --git a/litellm/llms/mistral/chat/transformation.py b/litellm/llms/mistral/chat/transformation.py index 15e348453e0..a402e2c4048 100644 --- a/litellm/llms/mistral/chat/transformation.py +++ b/litellm/llms/mistral/chat/transformation.py @@ -25,6 +25,7 @@ from litellm.litellm_core_utils.prompt_templates.common_utils import ( strip_litellm_internal_message_fields, strip_none_values_from_message, ) +from litellm.litellm_core_utils.reasoning_effort_utils import OPAQUE_VALUE, warn_dropped_reasoning from litellm.litellm_core_utils.url_utils import ensure_api_route from litellm.llms.openai.chat.gpt_transformation import ( OpenAIChatCompletionStreamingHandler, @@ -74,6 +75,12 @@ _MISTRAL_REQUEST_ROUTING: Final = TypeAdapter(MistralRequestRouting) _MISTRAL_CACHE_KEY_PARAMS: Final = TypeAdapter(MistralCacheKeyParams) _MISTRAL_REASONING_EFFORT_PARAMS: Final = TypeAdapter(MistralReasoningEffortParams) _MISTRAL_FILE_PART: Final = TypeAdapter(MistralFilePart) +_REASONING_PROMPT_REASON: Final = "litellm sends Magistral a reasoning system prompt instead of the requested level." +_NON_LEVEL_EFFORT_REASON: Final = ( + "the Mistral API takes it only as a level such as 'high', so Mistral applies its default effort." +) +_DOCUMENTED_MAGISTRAL_EFFORTS: Final = frozenset({"low", "medium", "high"}) +_REASONING_ON_THINKING_TYPES: Final = frozenset({"enabled", "adaptive"}) def _reasoning_level_and_summary(reasoning_effort: object) -> tuple[object, object]: @@ -83,6 +90,16 @@ def _reasoning_level_and_summary(reasoning_effort: object) -> tuple[object, obje return requested.get("effort"), requested.get("summary") +def _warn_unless_reasoning_prompt_honors(param: str, value: object, model: str) -> None: + if param == "reasoning_effort" and isinstance(value, str) and value in _DOCUMENTED_MAGISTRAL_EFFORTS: + return + thinking: Final = as_str_mapping(value) + thinking_type: Final = thinking.get("type") if thinking is not None else None + if param == "thinking" and isinstance(thinking_type, str) and thinking_type in _REASONING_ON_THINKING_TYPES: + return + warn_dropped_reasoning(param, value, model, _REASONING_PROMPT_REASON, drop_params_gated=False) + + def _is_json_array(value: object) -> TypeIs[Sequence[object]]: # guard-ok: a decoded JSON array is a list return isinstance(value, list) @@ -253,6 +270,7 @@ class MistralConfig(OpenAIGPTConfig): if param in ("reasoning_effort", "thinking") and "magistral" in model.lower(): # Flag that we need to add reasoning system prompt optional_params["_add_reasoning_prompt"] = True + _warn_unless_reasoning_prompt_honors(param, OPAQUE_VALUE.validate_python(value), model) if param in ("parallel_tool_calls", "prompt_cache_key"): optional_params[param] = value return optional_params @@ -260,13 +278,18 @@ class MistralConfig(OpenAIGPTConfig): def _reasoning_level_to_send( self, model: str, requested: MistralReasoningEffortParams, drop_params: bool ) -> str | None: - level, summary = _reasoning_level_and_summary(requested.get("reasoning_effort")) + requested_effort: Final = requested.get("reasoning_effort") + level, summary = _reasoning_level_and_summary(requested_effort) sendable: Final = ( _accepted_reasoning_effort(model, level, self.custom_llm_provider) if isinstance(level, str) else None ) if sendable is not None and summary is None: return sendable if drop_params or litellm.drop_params is True: + if sendable is None and level is not None: + warn_dropped_reasoning( + "reasoning_effort", requested_effort, model, _NON_LEVEL_EFFORT_REASON, drop_params_gated=True + ) return sendable raise UnsupportedParamsError( message=( diff --git a/litellm/llms/oci/chat/transformation.py b/litellm/llms/oci/chat/transformation.py index 8257a75b4a5..9c827822dc1 100644 --- a/litellm/llms/oci/chat/transformation.py +++ b/litellm/llms/oci/chat/transformation.py @@ -21,6 +21,7 @@ from litellm._logging import verbose_logger from litellm.constants import DEFAULT_OCI_CHAT_MAX_TOKENS from litellm.litellm_core_utils.logging_utils import track_llm_api_timing from litellm.litellm_core_utils.prompt_templates.common_utils import messages_with_file_parts_as_image_urls +from litellm.litellm_core_utils.reasoning_effort_utils import warn_dropped_reasoning from litellm.litellm_core_utils.streaming_handler import CustomStreamWrapper from litellm.litellm_core_utils.url_utils import join_api_base from litellm.llms.base_llm.chat.transformation import BaseConfig, BaseLLMException @@ -428,6 +429,13 @@ class OCIChatConfig(BaseConfig): alias = param_map.get(key) if key == "reasoning_effort" and alias and not _model_supports_reasoning_effort(model): if drop_params or litellm.drop_params: + warn_dropped_reasoning( + key, + value, + model, + "the model map lists no reasoning_effort levels for it, so OCI would reject reasoningEffort.", + drop_params_gated=True, + ) continue raise OCIError( status_code=400, @@ -451,6 +459,14 @@ class OCIChatConfig(BaseConfig): if key == "n" and (value is None or value == 1): continue if drop_params or litellm.drop_params: + if key == "reasoning_effort": + warn_dropped_reasoning( + key, + value, + model, + "OCI's Cohere chat API has no reasoning effort field.", + drop_params_gated=True, + ) continue raise OCIError( status_code=400, diff --git a/litellm/llms/ollama/chat/transformation.py b/litellm/llms/ollama/chat/transformation.py index f9e7071db41..755d647da1d 100644 --- a/litellm/llms/ollama/chat/transformation.py +++ b/litellm/llms/ollama/chat/transformation.py @@ -14,6 +14,7 @@ from litellm.litellm_core_utils.prompt_templates.common_utils import ( extract_images_from_message, extract_reasoning_content, ) +from litellm.litellm_core_utils.reasoning_effort_utils import OPAQUE_VALUE from litellm.litellm_core_utils.url_utils import ensure_api_route from litellm.llms.base_llm.base_model_iterator import BaseModelResponseIterator from litellm.llms.base_llm.chat.transformation import BaseConfig, BaseLLMException @@ -29,7 +30,7 @@ from litellm.types.llms.openai import ( ) from litellm.types.utils import ModelResponse, ModelResponseStream -from ..common_utils import OllamaError +from ..common_utils import OllamaError, ollama_think if TYPE_CHECKING: from litellm.litellm_core_utils.litellm_logging import Logging as _LiteLLMLoggingObj @@ -172,10 +173,7 @@ class OllamaChatConfig(BaseConfig): if value.get("json_schema") and value["json_schema"].get("schema"): optional_params["format"] = value["json_schema"]["schema"] if param == "reasoning_effort" and value is not None: - if model.startswith("gpt-oss"): - optional_params["think"] = value - else: - optional_params["think"] = value in {"low", "medium", "high"} + optional_params["think"] = ollama_think(model, OPAQUE_VALUE.validate_python(value)) ### FUNCTION CALLING LOGIC ### # Ollama 0.4+ supports native tool calling - pass tools directly # and let Ollama handle model capability detection diff --git a/litellm/llms/ollama/common_utils.py b/litellm/llms/ollama/common_utils.py index f313c62f026..fcf6a396a13 100644 --- a/litellm/llms/ollama/common_utils.py +++ b/litellm/llms/ollama/common_utils.py @@ -8,11 +8,14 @@ from pydantic import BaseModel, TypeAdapter from litellm import verbose_logger from litellm._uuid import uuid +from litellm.litellm_core_utils.reasoning_effort_utils import warn_dropped_reasoning from litellm.litellm_core_utils.url_utils import join_api_base from litellm.llms.base_llm.chat.transformation import BaseLLMException from litellm.types.utils import ModelResponse, ModelResponseStream _PROVIDER_FIELDS: Final = TypeAdapter(dict[str, object]) +_THINK_ON_EFFORTS: Final = frozenset({"low", "medium", "high"}) +_THINK_OFF_REASON: Final = "Ollama takes think only as on or off, and litellm maps this level to off." _DECODED_JSON: Final = TypeAdapter(object) @@ -296,3 +299,12 @@ class OllamaModelInfo(BaseLLMModelInfo): Return the base model name for Ollama (no-op). """ return model + + +def ollama_think(model: str, reasoning_effort: object) -> object: + if model.startswith("gpt-oss"): + return reasoning_effort + think: Final = reasoning_effort in _THINK_ON_EFFORTS + if isinstance(reasoning_effort, str) and not think and reasoning_effort != "none": + warn_dropped_reasoning("reasoning_effort", reasoning_effort, model, _THINK_OFF_REASON, drop_params_gated=False) + return think diff --git a/litellm/llms/ollama/completion/transformation.py b/litellm/llms/ollama/completion/transformation.py index 239b4e32645..6dbcb050e11 100644 --- a/litellm/llms/ollama/completion/transformation.py +++ b/litellm/llms/ollama/completion/transformation.py @@ -23,6 +23,7 @@ from litellm.litellm_core_utils.prompt_templates.image_handling import ( async_inline_remote_media, inline_remote_image_urls, ) +from litellm.litellm_core_utils.reasoning_effort_utils import OPAQUE_VALUE from litellm.litellm_core_utils.url_utils import ensure_api_route from litellm.llms.base_llm.base_model_iterator import BaseModelResponseIterator from litellm.llms.base_llm.chat.transformation import BaseConfig, BaseLLMException @@ -43,7 +44,7 @@ from litellm.types.utils import ( generate_id, ) -from ..common_utils import OllamaError, OllamaModelInfo, convert_image +from ..common_utils import OllamaError, OllamaModelInfo, convert_image, ollama_think if TYPE_CHECKING: from litellm.litellm_core_utils.litellm_logging import Logging as _LiteLLMLoggingObj @@ -249,10 +250,7 @@ class OllamaConfig(BaseConfig): elif param == "stop": optional_params["stop"] = value elif param == "reasoning_effort" and value is not None: - if model.startswith("gpt-oss"): - optional_params["think"] = value - else: - optional_params["think"] = value in {"low", "medium", "high"} + optional_params["think"] = ollama_think(model, OPAQUE_VALUE.validate_python(value)) elif param == "response_format" and isinstance(value, dict): if value["type"] == "json_object": optional_params["format"] = "json" diff --git a/litellm/llms/openai/common_utils.py b/litellm/llms/openai/common_utils.py index aa14a985afe..c113cb351d8 100644 --- a/litellm/llms/openai/common_utils.py +++ b/litellm/llms/openai/common_utils.py @@ -32,6 +32,7 @@ import litellm from litellm.exceptions import error_response_without_credentials from litellm.litellm_core_utils.completion_timeout import CompletionTimeout from litellm.litellm_core_utils.core_helpers import as_str_mapping, decoded_json_object +from litellm.litellm_core_utils.reasoning_effort_utils import warn_dropped_reasoning_params from litellm.litellm_core_utils.token_counter import token_counter, usage_tokenizer_model from litellm.llms.base_llm.chat.transformation import BaseLLMException from litellm.llms.custom_httpx.http_handler import ( @@ -165,17 +166,11 @@ _DECODED_JSON: Final = TypeAdapter(object) def drop_params_from_unprocessable_entity_error( e: openai.UnprocessableEntityError | httpx.HTTPStatusError, data: Mapping[str, object], + *, + model: str | None, + drop_params_gated: bool, ) -> dict[str, object]: - """ - Helper function to read OpenAI UnprocessableEntityError and drop the params that raised an error from the error message. - - Args: - e (UnprocessableEntityError): The UnprocessableEntityError exception - data (Dict[str, Any]): The original data dictionary containing all parameters - - Returns: - Dict[str, Any]: A new dictionary with invalid parameters removed - """ + """Return a copy of ``data`` without the params a 422 response names as invalid.""" invalid_params: Final[list[str]] = [] if isinstance(e, httpx.HTTPStatusError): error_json: Final = decoded_json_object(_DECODED_JSON.validate_python(e.response.json())) @@ -203,6 +198,13 @@ def drop_params_from_unprocessable_entity_error( invalid_params.append(error_dict["loc"][1]) new_data: Final = {k: v for k, v in data.items() if k not in invalid_params} + warn_dropped_reasoning_params( + invalid_params, + data, + model, + "the provider answered 422 naming it an invalid param.", + drop_params_gated=drop_params_gated, + ) return new_data diff --git a/litellm/llms/openai/openai.py b/litellm/llms/openai/openai.py index bef99fd0719..bdd9f285823 100644 --- a/litellm/llms/openai/openai.py +++ b/litellm/llms/openai/openai.py @@ -940,7 +940,9 @@ class OpenAIChatCompletion(BaseLLM, BaseOpenAILLM): except openai.UnprocessableEntityError as e: ## check if body contains unprocessable params - related issue https://github.com/BerriAI/litellm/issues/4800 if litellm.drop_params is True or drop_params is True: - inference_params = drop_params_from_unprocessable_entity_error(e, inference_params) + inference_params = drop_params_from_unprocessable_entity_error( + e, inference_params, model=model, drop_params_gated=True + ) else: raise e # e.message @@ -1097,7 +1099,7 @@ class OpenAIChatCompletion(BaseLLM, BaseOpenAILLM): except openai.UnprocessableEntityError as e: ## check if body contains unprocessable params - related issue https://github.com/BerriAI/litellm/issues/4800 if litellm.drop_params is True or drop_params is True: - data = drop_params_from_unprocessable_entity_error(e, data) + data = drop_params_from_unprocessable_entity_error(e, data, model=model, drop_params_gated=True) else: raise e # e.message @@ -1252,7 +1254,7 @@ class OpenAIChatCompletion(BaseLLM, BaseOpenAILLM): except openai.UnprocessableEntityError as e: ## check if body contains unprocessable params - related issue https://github.com/BerriAI/litellm/issues/4800 if litellm.drop_params is True or drop_params is True: - data = drop_params_from_unprocessable_entity_error(e, data) + data = drop_params_from_unprocessable_entity_error(e, data, model=model, drop_params_gated=True) else: raise e except ( diff --git a/litellm/llms/tencent/chat/transformation.py b/litellm/llms/tencent/chat/transformation.py index 1f0bcd218d6..23a6ccd52c6 100644 --- a/litellm/llms/tencent/chat/transformation.py +++ b/litellm/llms/tencent/chat/transformation.py @@ -9,12 +9,15 @@ from typing import Final, TypedDict from typing_extensions import ReadOnly import litellm +from litellm.litellm_core_utils.reasoning_effort_utils import warn_dropped_reasoning from litellm.litellm_core_utils.url_utils import ensure_api_route, join_api_base from litellm.secret_managers.main import get_secret_str from litellm.utils import supports_reasoning from ...openai.chat.gpt_transformation import OpenAIGPTConfig +_ADAPTIVE_BUDGET_REASON: Final = "it is not an integer, so litellm sends adaptive thinking with no budget." + class ThinkingPayload(TypedDict, total=False): """Tencent TokenHub `thinking` object. @@ -102,6 +105,10 @@ class TencentChatConfig(OpenAIGPTConfig): if isinstance(budget, int): coerced_with_budget: Final[ThinkingPayload] = {"type": "adaptive", "budget_tokens": budget} return coerced_with_budget + if budget is not None: + warn_dropped_reasoning( + "thinking.budget_tokens", budget, model, _ADAPTIVE_BUDGET_REASON, drop_params_gated=False + ) coerced: Final[ThinkingPayload] = {"type": "adaptive"} return coerced diff --git a/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py b/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py index c506192f253..1ef5267df14 100644 --- a/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py +++ b/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py @@ -34,6 +34,7 @@ from litellm.litellm_core_utils.json_fragment_accumulator import JSONFragmentAcc from litellm.litellm_core_utils.prompt_templates.factory import ( encode_tool_call_id_with_signature, ) +from litellm.litellm_core_utils.reasoning_effort_utils import warn_dropped_reasoning from litellm.llms.base_llm.chat.transformation import BaseConfig, BaseLLMException from litellm.llms.custom_httpx.http_handler import ( AsyncHTTPHandler, @@ -201,6 +202,13 @@ def _unsupported_reasoning_effort(reasoning_effort: str) -> UnsupportedParamsErr _SAMPLING_DEPRECATION_UNMAPPED_MODELS: Final = "models outside litellm's model map" +_NON_INTEGER_BUDGET_REASON: Final = ( + "Gemini takes budget_tokens only as an integer, so it uses its default thinking budget." +) +_LEVEL_ONLY_BUDGET_REASON: Final = ( + "Gemini 3 and later set reasoning by thinking level, not budget, so litellm does not send the budget. " + "Send reasoning_effort to pick a level." +) _sampling_deprecation_lock: Final = threading.Lock() @@ -1466,6 +1474,14 @@ class VertexGeminiConfig(VertexAIBaseConfig, BaseConfig): params["includeThoughts"] = False else: params["includeThoughts"] = sends_thoughts + if thinking_budget is not None: + warn_dropped_reasoning( + "thinking.budget_tokens", + thinking_budget, + model, + _LEVEL_ONLY_BUDGET_REASON, + drop_params_gated=False, + ) # Follow provider defaults unless explicitly opted into legacy behavior. if litellm.enable_gemini_default_thinking_level_low is True: params["thinkingLevel"] = ( @@ -1478,8 +1494,16 @@ class VertexGeminiConfig(VertexAIBaseConfig, BaseConfig): # For older Gemini models, use thinkingBudget if thinking_enabled and not VertexGeminiConfig._is_thinking_budget_zero(thinking_budget): params["includeThoughts"] = sends_thoughts - if thinking_budget is not None and isinstance(thinking_budget, int): + if isinstance(thinking_budget, int): params["thinkingBudget"] = thinking_budget + elif thinking_budget is not None: + warn_dropped_reasoning( + "thinking.budget_tokens", + thinking_budget, + model, + _NON_INTEGER_BUDGET_REASON, + drop_params_gated=False, + ) return params diff --git a/litellm/router_utils/reasoning_effort_capability.py b/litellm/router_utils/reasoning_effort_capability.py index 3563fcd96ae..c315a31b65d 100644 --- a/litellm/router_utils/reasoning_effort_capability.py +++ b/litellm/router_utils/reasoning_effort_capability.py @@ -38,7 +38,7 @@ from litellm.litellm_core_utils.reasoning_effort_strength import ( ReasoningEffortRounding, nearest_declared_reasoning_effort, ) -from litellm.litellm_core_utils.reasoning_effort_utils import reasoning_effort_from_thinking +from litellm.litellm_core_utils.reasoning_effort_utils import reasoning_effort_from_thinking_silently from litellm.llms.anthropic.common_utils import AnthropicModelInfo, validated_object_mapping from litellm.types.llms.anthropic import AnthropicThinkingParam from litellm.types.llms.openai import REASONING_EFFORT @@ -143,7 +143,7 @@ def with_adaptive_thinking_as_effort( params: Mapping[str, object], *, model: str, custom_llm_provider: str, drop_params: bool ) -> Mapping[str, object]: translated: Final = dict(params) - AnthropicModelInfo.translate_legacy_thinking_for_adaptive_model( + AnthropicModelInfo.translate_legacy_thinking_for_adaptive_model_silently( model=model, optional_params=translated, custom_llm_provider=custom_llm_provider, drop_params=drop_params ) return translated @@ -151,7 +151,7 @@ def with_adaptive_thinking_as_effort( def thinking_budget_effort(thinking: object, *, model: str, custom_llm_provider: str, drop_params: bool) -> str | None: if custom_llm_provider in _PROVIDERS_SENDING_THINKING_AS_EFFORT: - return reasoning_effort_from_thinking(thinking, drop_params=drop_params) + return reasoning_effort_from_thinking_silently(thinking, drop_params=drop_params) adaptive_config: Final = as_str_mapping( with_adaptive_thinking_as_effort( {"thinking": thinking}, model=model, custom_llm_provider=custom_llm_provider, drop_params=drop_params @@ -162,7 +162,7 @@ def thinking_budget_effort(thinking: object, *, model: str, custom_llm_provider: return adaptive_effort if isinstance(adaptive_effort, str) else None if "thinking" in (get_supported_openai_params(model=model, custom_llm_provider=custom_llm_provider) or ()): return None - return reasoning_effort_from_thinking(thinking, drop_params=drop_params) + return reasoning_effort_from_thinking_silently(thinking, drop_params=drop_params) def provider_reads_thinking(thinking: object, *, model: str, custom_llm_provider: str, drop_params: bool) -> bool: diff --git a/tests/unit/llms/anthropic/chat/guardrail_translation/test_anthropic_guardrail_handler.py b/tests/unit/llms/anthropic/chat/guardrail_translation/test_anthropic_guardrail_handler.py index d815e1eedd3..c2ee58b67ed 100644 --- a/tests/unit/llms/anthropic/chat/guardrail_translation/test_anthropic_guardrail_handler.py +++ b/tests/unit/llms/anthropic/chat/guardrail_translation/test_anthropic_guardrail_handler.py @@ -6,6 +6,7 @@ with guardrail transformations, specifically testing edge cases with empty choic """ import json +import logging import re from typing import Any, Literal, Optional, Final from unittest.mock import MagicMock, patch @@ -3391,3 +3392,18 @@ class TestAnthropicMessagesHandlerPostCallHookResponse: native = {"type": "message", "role": "assistant", "content": [{"type": "text", "text": "hi"}]} assert AnthropicMessagesHandler().post_call_hook_response(native) is native + + +def test_reading_a_dropped_budget_for_a_guardrail_logs_no_drop_warning(caplog: pytest.LogCaptureFixture) -> None: + data: Final = { + "model": "gemini-2.5-flash", + "max_tokens": 64, + "thinking": {"type": "enabled", "budget_tokens": "high"}, + "messages": [{"role": "user", "content": "hi"}], + } + + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + messages: Final = AnthropicMessagesHandler().get_structured_messages(data) + + assert messages == [{"role": "user", "content": "hi"}] + assert not [record for record in caplog.records if "Dropping" in record.getMessage()] diff --git a/tests/unit/llms/anthropic/chat/test_anthropic_chat_transformation.py b/tests/unit/llms/anthropic/chat/test_anthropic_chat_transformation.py index 4d5d6cc17a9..cbb60caa57b 100644 --- a/tests/unit/llms/anthropic/chat/test_anthropic_chat_transformation.py +++ b/tests/unit/llms/anthropic/chat/test_anthropic_chat_transformation.py @@ -1,5 +1,6 @@ import copy import json +import logging from itertools import chain from typing import Final from unittest.mock import MagicMock, patch @@ -3189,6 +3190,47 @@ def test_reasoning_effort_accepts_dict_shape_for_non_adaptive_model( ) +def _reasoning_effort_drop_warnings(caplog: pytest.LogCaptureFixture) -> tuple[str, ...]: + return tuple( + record.getMessage() for record in caplog.records if "Dropping reasoning_effort=" in record.getMessage() + ) + + +@pytest.mark.parametrize("reasoning_effort_value", [5, {"effort": 5, "summary": "concise"}]) +def test_reasoning_effort_that_is_not_a_level_name_is_dropped_with_one_warning( + reasoning_effort_value: object, caplog: pytest.LogCaptureFixture +) -> None: + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + result: Final = AnthropicConfig().map_openai_params( + non_default_params={"reasoning_effort": reasoning_effort_value}, + optional_params={}, + model="claude-sonnet-4-6-20260219", + drop_params=False, + ) + + warnings: Final = _reasoning_effort_drop_warnings(caplog) + assert "thinking" not in result + assert "output_config" not in result + assert len(warnings) == 1 + assert f"reasoning_effort={reasoning_effort_value!r}" in warnings[0] + assert "claude-sonnet-4-6-20260219" in warnings[0] + + +@pytest.mark.parametrize("reasoning_effort_value", [{"summary": "concise"}, {"effort": "low", "summary": "concise"}]) +def test_reasoning_effort_with_a_level_name_or_no_level_logs_no_drop( + reasoning_effort_value: object, caplog: pytest.LogCaptureFixture +) -> None: + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + AnthropicConfig().map_openai_params( + non_default_params={"reasoning_effort": reasoning_effort_value}, + optional_params={}, + model="claude-sonnet-4-6-20260219", + drop_params=False, + ) + + assert _reasoning_effort_drop_warnings(caplog) == () + + @pytest.mark.parametrize( "model,budget_tokens,expected", [ diff --git a/tests/unit/llms/anthropic/messages/adapters/test_anthropic_messages_adapters_transformation.py b/tests/unit/llms/anthropic/messages/adapters/test_anthropic_messages_adapters_transformation.py index c30644e7fdc..e0882653324 100644 --- a/tests/unit/llms/anthropic/messages/adapters/test_anthropic_messages_adapters_transformation.py +++ b/tests/unit/llms/anthropic/messages/adapters/test_anthropic_messages_adapters_transformation.py @@ -1,6 +1,7 @@ import base64 import functools import json +import logging from collections.abc import Mapping from types import MappingProxyType from typing import Any, Final, cast @@ -2871,14 +2872,21 @@ def test_translate_completion_input_params_rejects_a_non_numeric_budget_without_ @pytest.mark.parametrize("budget", ["2048", True]) def test_translate_completion_input_params_drops_a_non_numeric_budget_under_drop_params( - budget: object, monkeypatch: pytest.MonkeyPatch + budget: object, monkeypatch: pytest.MonkeyPatch, caplog: pytest.LogCaptureFixture ): monkeypatch.setattr(litellm, "drop_params", False) - translated: Final = AnthropicAdapter().translate_completion_input_params( - _gemini_request_with_budget(budget, drop_params=True) - ) + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + translated: Final = AnthropicAdapter().translate_completion_input_params( + _gemini_request_with_budget(budget, drop_params=True) + ) + drop_warnings: Final = [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] assert translated is not None assert not {"reasoning_effort", "thinking"} & translated.keys() + assert len(drop_warnings) == 1 + assert all( + fragment in drop_warnings[0] + for fragment in (f"thinking.budget_tokens={budget!r}", "gemini-2.5-flash", "(drop_params=True)") + ) def test_translate_openai_content_to_anthropic_reasoning_content_without_thinking_blocks(): diff --git a/tests/unit/llms/anthropic/messages/adapters/test_handler_reasoning_effort_normalization.py b/tests/unit/llms/anthropic/messages/adapters/test_handler_reasoning_effort_normalization.py index e504513dd36..27ed1f5ebd6 100644 --- a/tests/unit/llms/anthropic/messages/adapters/test_handler_reasoning_effort_normalization.py +++ b/tests/unit/llms/anthropic/messages/adapters/test_handler_reasoning_effort_normalization.py @@ -8,6 +8,7 @@ leaves the adapter, in the shape the target expects. import functools import json +import logging from collections.abc import Mapping from typing import Final @@ -510,26 +511,33 @@ class TestADeclaredDeploymentRoundsTheMessagesTierOntoItsLevels: @pytest.mark.asyncio @pytest.mark.parametrize("budget", ["2048", True]) @pytest.mark.parametrize("litellm_params, model_info, upstream_url, upstream_body", _MALFORMED_BUDGET_TARGETS) - async def test_drop_params_drops_a_non_numeric_budget( + async def test_drop_params_drops_a_non_numeric_budget_with_one_warning( self, local_model_cost_map: None, respx_mock: respx.MockRouter, + caplog: pytest.LogCaptureFixture, litellm_params: Mapping[str, object], model_info: Mapping[str, object], upstream_url: str, upstream_body: Mapping[str, object], budget: object, ) -> None: - sent: Final = await _body_sent_upstream( - respx_mock, - {**litellm_params, "drop_params": True}, - model_info, - upstream_url, - upstream_body, - thinking={"type": "enabled", "budget_tokens": budget}, - ) + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + sent: Final = await _body_sent_upstream( + respx_mock, + {**litellm_params, "drop_params": True}, + model_info, + upstream_url, + upstream_body, + thinking={"type": "enabled", "budget_tokens": budget}, + ) assert not {"reasoning_effort", "reasoning", "thinking"} & sent.keys() + dropped: Final = [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] + assert len(dropped) == 1 + assert f"thinking.budget_tokens={budget!r}" in dropped[0] + assert "(drop_params=True)" in dropped[0] + assert str(litellm_params["model"]).split("/", 1)[1] in dropped[0] @pytest.mark.asyncio @pytest.mark.parametrize( diff --git a/tests/unit/llms/anthropic/messages/native/test_anthropic_messages_handler.py b/tests/unit/llms/anthropic/messages/native/test_anthropic_messages_handler.py index 5ce6d03f477..af3108e5072 100644 --- a/tests/unit/llms/anthropic/messages/native/test_anthropic_messages_handler.py +++ b/tests/unit/llms/anthropic/messages/native/test_anthropic_messages_handler.py @@ -159,7 +159,7 @@ async def test_openai_model_does_not_forward_stream_options_to_responses_api(): await litellm.anthropic.messages.acreate( max_tokens=100, messages=[{"role": "user", "content": "Hello, how are you?"}], - model="openai/gpt-5.5", + model="openai/gpt-6.1-sol", api_key="test-api-key", stream_options={"include_usage": True}, ) @@ -669,7 +669,9 @@ class TestThinkingSummaryPreservation: ) thinking = {"type": "enabled", "budget_tokens": 5000, "summary": "concise"} - result = LiteLLMAnthropicToResponsesAPIAdapter.translate_thinking_to_reasoning(thinking) + result = LiteLLMAnthropicToResponsesAPIAdapter.translate_thinking_to_reasoning( + thinking, model="openai/gpt-6.1-sol" + ) assert result == {"effort": "high", "summary": "concise"} def test_responses_adapter_no_summary_by_default(self): @@ -683,7 +685,9 @@ class TestThinkingSummaryPreservation: try: litellm.reasoning_auto_summary = False thinking = {"type": "enabled", "budget_tokens": 5000} - result = LiteLLMAnthropicToResponsesAPIAdapter.translate_thinking_to_reasoning(thinking) + result = LiteLLMAnthropicToResponsesAPIAdapter.translate_thinking_to_reasoning( + thinking, model="openai/gpt-6.1-sol" + ) assert result == {"effort": "high"} assert result is not None and "summary" not in result finally: diff --git a/tests/unit/llms/anthropic/messages/native/test_anthropic_messages_router.py b/tests/unit/llms/anthropic/messages/native/test_anthropic_messages_router.py index 596cb968a0e..1d0d70d8e60 100644 --- a/tests/unit/llms/anthropic/messages/native/test_anthropic_messages_router.py +++ b/tests/unit/llms/anthropic/messages/native/test_anthropic_messages_router.py @@ -3,6 +3,7 @@ from __future__ import annotations import asyncio import base64 import json +import logging import struct import time import uuid @@ -891,24 +892,32 @@ async def test_native_messages_send_no_effort_for_a_default_effort_beside_a_thin @pytest.mark.asyncio @pytest.mark.parametrize("model_info", [None, _DECLARING_LOW_AND_HIGH]) @pytest.mark.parametrize("drop_globally", [False, True]) -async def test_native_messages_drop_params_drops_a_non_numeric_thinking_budget_on_adaptive_claude( +async def test_native_messages_drop_params_drops_a_non_numeric_thinking_budget_on_adaptive_claude_with_one_warning( respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch, + caplog: pytest.LogCaptureFixture, model_info: Mapping[str, object] | None, drop_globally: bool, ) -> None: monkeypatch.setattr(litellm, "drop_params", drop_globally) request_drop: Final = {} if drop_globally else {"drop_params": True} - dropped: Final = await _native_opus_body_sent( - respx_mock, model_info, thinking={"type": "enabled", "budget_tokens": "lots"}, **request_drop - ) + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + dropped: Final = await _native_opus_body_sent( + respx_mock, model_info, thinking={"type": "enabled", "budget_tokens": "lots"}, **request_drop + ) + drop_warnings: Final = [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] adaptive_alone: Final = await _native_opus_body_sent( respx_mock, model_info, thinking={"type": "adaptive"}, **request_drop ) assert "output_config" not in adaptive_alone assert dropped == adaptive_alone + assert len(drop_warnings) == 1 + assert all( + fragment in drop_warnings[0] + for fragment in ("thinking.budget_tokens='lots'", "claude-opus-5-5", "(drop_params=True)") + ) _MIXED_GROUP: Final = "gpt-5.6-mixed" diff --git a/tests/unit/llms/anthropic/messages/native/test_reasoning_effort_translation.py b/tests/unit/llms/anthropic/messages/native/test_reasoning_effort_translation.py index 50dc0deabfd..28abb348371 100644 --- a/tests/unit/llms/anthropic/messages/native/test_reasoning_effort_translation.py +++ b/tests/unit/llms/anthropic/messages/native/test_reasoning_effort_translation.py @@ -1,5 +1,8 @@ """Tests for ``reasoning_effort`` translation on the Anthropic /v1/messages route.""" +import logging +from typing import Final + import pytest from litellm.constants import ( @@ -46,6 +49,49 @@ def test_reasoning_effort_maps_to_output_config_for_adaptive_model( assert result.get("output_config") == {"effort": expected_effort} +def _reasoning_effort_drop_warnings(caplog: pytest.LogCaptureFixture) -> tuple[str, ...]: + return tuple( + record.getMessage() for record in caplog.records if "Dropping reasoning_effort=" in record.getMessage() + ) + + +@pytest.mark.parametrize("reasoning_effort", [5, {"effort": "high"}]) +def test_reasoning_effort_that_is_not_a_string_is_dropped_with_one_warning( + reasoning_effort: object, caplog: pytest.LogCaptureFixture +) -> None: + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + result: Final = AnthropicMessagesConfig().transform_anthropic_messages_request( + model="claude-opus-4-7", + messages=[{"role": "user", "content": "Hello"}], + anthropic_messages_optional_request_params={"max_tokens": 1024, "reasoning_effort": reasoning_effort}, + litellm_params={}, + headers={}, + ) + + warnings: Final = _reasoning_effort_drop_warnings(caplog) + assert "reasoning_effort" not in result + assert "output_config" not in result + assert len(warnings) == 1 + assert f"reasoning_effort={reasoning_effort!r}" in warnings[0] + assert "claude-opus-4-7" in warnings[0] + + +@pytest.mark.parametrize("reasoning_effort", ["high", {"summary": "concise"}]) +def test_reasoning_effort_that_is_a_string_or_names_no_level_logs_no_drop( + reasoning_effort: object, caplog: pytest.LogCaptureFixture +) -> None: + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + AnthropicMessagesConfig().transform_anthropic_messages_request( + model="claude-opus-4-7", + messages=[{"role": "user", "content": "Hello"}], + anthropic_messages_optional_request_params={"max_tokens": 1024, "reasoning_effort": reasoning_effort}, + litellm_params={}, + headers={}, + ) + + assert _reasoning_effort_drop_warnings(caplog) == () + + def test_reasoning_effort_none_clears_thinking_and_output_config(): config = AnthropicMessagesConfig() optional_params = { diff --git a/tests/unit/llms/anthropic/messages/responses_adapters/test_responses_adapters_transformation.py b/tests/unit/llms/anthropic/messages/responses_adapters/test_responses_adapters_transformation.py index f6fcb9e6f1e..ece0cf5d423 100644 --- a/tests/unit/llms/anthropic/messages/responses_adapters/test_responses_adapters_transformation.py +++ b/tests/unit/llms/anthropic/messages/responses_adapters/test_responses_adapters_transformation.py @@ -44,6 +44,7 @@ def _make_request(**overrides) -> AnthropicMessagesRequest: _ADAPTER = LiteLLMAnthropicToResponsesAPIAdapter() +_MODEL: Final = "openai/gpt-6.1-sol" # --------------------------------------------------------------------------- @@ -957,13 +958,13 @@ class TestTranslateThinkingToReasoning: """Anthropic thinking param -> Responses API reasoning param.""" def test_budget_high_effort(self): - result = _ADAPTER.translate_thinking_to_reasoning({"type": "enabled", "budget_tokens": 10000}) + result = _ADAPTER.translate_thinking_to_reasoning({"type": "enabled", "budget_tokens": 10000}, model=_MODEL) # Default (reasoning_auto_summary=False): only effort, no summary assert result == {"effort": "high"} assert result is not None and "summary" not in result def test_budget_above_threshold_high_effort(self): - result = _ADAPTER.translate_thinking_to_reasoning({"type": "enabled", "budget_tokens": 50000}) + result = _ADAPTER.translate_thinking_to_reasoning({"type": "enabled", "budget_tokens": 50000}, model=_MODEL) assert result is not None assert result["effort"] == "high" assert "summary" not in result @@ -973,7 +974,8 @@ class TestTranslateThinkingToReasoning: { "type": "enabled", "budget_tokens": DEFAULT_REASONING_EFFORT_MEDIUM_THINKING_BUDGET, - } + }, + model=_MODEL, ) assert result == {"effort": "medium"} assert result is not None and "summary" not in result @@ -983,13 +985,14 @@ class TestTranslateThinkingToReasoning: { "type": "enabled", "budget_tokens": DEFAULT_REASONING_EFFORT_LOW_THINKING_BUDGET, - } + }, + model=_MODEL, ) assert result == {"effort": "low"} assert result is not None and "summary" not in result def test_budget_minimal_effort(self): - result = _ADAPTER.translate_thinking_to_reasoning({"type": "enabled", "budget_tokens": 500}) + result = _ADAPTER.translate_thinking_to_reasoning({"type": "enabled", "budget_tokens": 500}, model=_MODEL) assert result == {"effort": "minimal"} assert result is not None and "summary" not in result @@ -998,7 +1001,8 @@ class TestTranslateThinkingToReasoning: { "type": "enabled", "budget_tokens": DEFAULT_REASONING_EFFORT_HIGH_THINKING_BUDGET, - } + }, + model=_MODEL, ) assert result_high is not None assert result_high["effort"] == "high" @@ -1006,7 +1010,8 @@ class TestTranslateThinkingToReasoning: { "type": "enabled", "budget_tokens": DEFAULT_REASONING_EFFORT_MEDIUM_THINKING_BUDGET, - } + }, + model=_MODEL, ) assert result_medium is not None assert result_medium["effort"] == "medium" @@ -1015,23 +1020,24 @@ class TestTranslateThinkingToReasoning: { "type": "enabled", "budget_tokens": DEFAULT_REASONING_EFFORT_LOW_THINKING_BUDGET, - } + }, + model=_MODEL, ) assert result_low is not None assert result_low["effort"] == "low" assert "summary" not in result_low def test_disabled_type_returns_none(self): - result = _ADAPTER.translate_thinking_to_reasoning({"type": "disabled"}) + result = _ADAPTER.translate_thinking_to_reasoning({"type": "disabled"}, model=_MODEL) assert result is None def test_non_dict_returns_none(self): - result = _ADAPTER.translate_thinking_to_reasoning("enabled") # type: ignore + result = _ADAPTER.translate_thinking_to_reasoning("enabled", model=_MODEL) # type: ignore assert result is None def test_missing_budget_defaults_to_minimal(self): """Missing budget_tokens defaults to 0, below the low threshold -> minimal.""" - result = _ADAPTER.translate_thinking_to_reasoning({"type": "enabled"}) + result = _ADAPTER.translate_thinking_to_reasoning({"type": "enabled"}, model=_MODEL) assert result == {"effort": "minimal"} assert result is not None and "summary" not in result @@ -1042,7 +1048,7 @@ class TestTranslateThinkingToReasoning: original = litellm.reasoning_auto_summary try: litellm.reasoning_auto_summary = True - result = _ADAPTER.translate_thinking_to_reasoning({"type": "enabled", "budget_tokens": 10000}) + result = _ADAPTER.translate_thinking_to_reasoning({"type": "enabled", "budget_tokens": 10000}, model=_MODEL) assert result == {"effort": "high", "summary": "detailed"} finally: litellm.reasoning_auto_summary = original @@ -1059,7 +1065,8 @@ class TestTranslateThinkingToReasoning: { "type": "enabled", "budget_tokens": DEFAULT_REASONING_EFFORT_MEDIUM_THINKING_BUDGET, - } + }, + model=_MODEL, ) assert result == {"effort": "medium", "summary": "detailed"} finally: diff --git a/tests/unit/llms/anthropic/messages/test_reasoning_effort_fields.py b/tests/unit/llms/anthropic/messages/test_reasoning_effort_fields.py index 3432c9503de..7128c04a056 100644 --- a/tests/unit/llms/anthropic/messages/test_reasoning_effort_fields.py +++ b/tests/unit/llms/anthropic/messages/test_reasoning_effort_fields.py @@ -162,7 +162,8 @@ class TestAdapterAdaptiveThinking: adapter = LiteLLMAnthropicMessagesAdapter() result = adapter.translate_anthropic_thinking_to_reasoning_effort( - {"type": "adaptive"} + {"type": "adaptive"}, + model="openai/gpt-6.1-sol", ) assert result == "medium" @@ -198,6 +199,7 @@ class TestAdapterAdaptiveThinking: result = LiteLLMAnthropicToResponsesAPIAdapter.translate_thinking_to_reasoning( thinking={"type": "adaptive"}, output_config={"effort": "xhigh"}, + model="openai/gpt-6.1-sol", ) assert result is not None assert result["effort"] == "xhigh" @@ -210,6 +212,7 @@ class TestAdapterAdaptiveThinking: result = LiteLLMAnthropicToResponsesAPIAdapter.translate_thinking_to_reasoning( thinking={"type": "adaptive"}, + model="openai/gpt-6.1-sol", ) assert result is not None assert result["effort"] == "medium" diff --git a/tests/unit/llms/anthropic/test_anthropic_common_utils.py b/tests/unit/llms/anthropic/test_anthropic_common_utils.py index 3ec72f29d34..c1f28e33969 100644 --- a/tests/unit/llms/anthropic/test_anthropic_common_utils.py +++ b/tests/unit/llms/anthropic/test_anthropic_common_utils.py @@ -11,6 +11,7 @@ Verifies that: """ import json +import logging import os import sys import threading @@ -4530,16 +4531,25 @@ def test_shared_legacy_thinking_translation_rejects_a_non_numeric_budget(budget_ @pytest.mark.usefixtures("local_model_cost_map") @pytest.mark.parametrize("budget_tokens", _NON_NUMERIC_BUDGETS) -def test_shared_legacy_thinking_translation_drops_a_non_numeric_budget_under_drop_params(budget_tokens: object) -> None: +def test_shared_legacy_thinking_translation_drops_a_non_numeric_budget_under_drop_params( + budget_tokens: object, caplog: pytest.LogCaptureFixture +) -> None: from litellm.llms.anthropic.common_utils import AnthropicModelInfo optional_params: Final = {"thinking": {"type": "enabled", "budget_tokens": budget_tokens}} - AnthropicModelInfo.translate_legacy_thinking_for_adaptive_model( - model="claude-opus-5", optional_params=optional_params, custom_llm_provider="azure_ai", drop_params=True - ) + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + AnthropicModelInfo.translate_legacy_thinking_for_adaptive_model( + model="claude-opus-5", optional_params=optional_params, custom_llm_provider="azure_ai", drop_params=True + ) + drop_warnings: Final = [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] assert optional_params == {"thinking": {"type": "adaptive"}} + assert len(drop_warnings) == 1 + assert all( + fragment in drop_warnings[0] + for fragment in (f"thinking.budget_tokens={budget_tokens!r}", "claude-opus-5", "(drop_params=True)") + ) @pytest.mark.parametrize( diff --git a/tests/unit/llms/anthropic/test_anthropic_prompt_cache_prediction.py b/tests/unit/llms/anthropic/test_anthropic_prompt_cache_prediction.py index 4d0ebef4f2d..d95f831df3e 100644 --- a/tests/unit/llms/anthropic/test_anthropic_prompt_cache_prediction.py +++ b/tests/unit/llms/anthropic/test_anthropic_prompt_cache_prediction.py @@ -1,4 +1,5 @@ import json +import logging from collections.abc import Mapping from datetime import datetime from types import SimpleNamespace @@ -597,6 +598,45 @@ async def test_native_count_preserves_settings_and_requires_every_prefix( assert unexpected.call_count == 0 +@pytest.mark.usefixtures("local_model_cost_map") +def test_native_baseline_body_drops_a_non_numeric_budget_without_logging_a_drop( + monkeypatch: pytest.MonkeyPatch, caplog: pytest.LogCaptureFixture +) -> None: + monkeypatch.setattr(litellm, "drop_params", True) + request: Final = { + "model": "claude-opus-5", + "max_tokens": 64, + "messages": [{"role": "user", "content": "hi"}], + "thinking": {"type": "enabled", "budget_tokens": "high"}, + } + + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + body: Final = prepare_native_baseline_body(request, "anthropic/claude-opus-5") + + assert body is not None + assert body["thinking"] == {"type": "adaptive"} + assert [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] == [] + + +@pytest.mark.usefixtures("local_model_cost_map") +def test_native_baseline_body_drops_a_non_string_reasoning_effort_without_logging_a_drop( + caplog: pytest.LogCaptureFixture, +) -> None: + request: Final = { + "model": "claude-opus-5", + "max_tokens": 64, + "messages": [{"role": "user", "content": "hi"}], + "reasoning_effort": 5, + } + + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + body: Final = prepare_native_baseline_body(request, "anthropic/claude-opus-5") + + assert body is not None + assert "reasoning_effort" not in body + assert [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] == [] + + def test_native_baseline_body_matches_the_sent_body_when_drop_params_meets_an_auto_added_beta( monkeypatch: pytest.MonkeyPatch, ) -> None: diff --git a/tests/unit/llms/azure_ai/chat/test_azure_ai_transformation.py b/tests/unit/llms/azure_ai/chat/test_azure_ai_transformation.py index 145d45317c3..9f68a4fb73f 100644 --- a/tests/unit/llms/azure_ai/chat/test_azure_ai_transformation.py +++ b/tests/unit/llms/azure_ai/chat/test_azure_ai_transformation.py @@ -1,4 +1,5 @@ import json +import logging import traceback from collections.abc import Callable from types import MappingProxyType @@ -758,6 +759,66 @@ def test_drop_tool_level_extra_fields_strips_copilot_mcp_server_name(): assert result["tools"][1]["function"]["name"] == "read_file" +def test_extra_parameters_error_drops_reasoning_effort_with_one_warning(caplog: pytest.LogCaptureFixture) -> None: + import httpx + + error_text: Final = json.dumps( + { + "error": { + "message": "Extra parameters ['reasoning_effort'] are not allowed when extra-parameters is not set " + "or set to be 'error'. Set extra-parameters to 'pass-through' to pass to the model." + } + } + ) + error: Final = httpx.HTTPStatusError( + message="422", + request=httpx.Request("POST", "https://example.services.ai.azure.com/models/chat/completions"), + response=httpx.Response(422, text=error_text), + ) + request_data: Final = { + "model": "Kimi-K2.6", + "messages": [{"role": "user", "content": "Say hi."}], + "reasoning_effort": "high", + } + + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + retried: Final = AzureAIStudioConfig().transform_request_on_unprocessable_entity_error(error, request_data) + drop_warnings: Final = [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] + + assert retried == {"model": "Kimi-K2.6", "messages": [{"role": "user", "content": "Say hi."}]} + assert len(drop_warnings) == 1 + assert all(fragment in drop_warnings[0] for fragment in ("reasoning_effort='high'", "Kimi-K2.6")) + assert "drop_params" not in drop_warnings[0] + + +def test_extra_inputs_error_names_drop_params_in_the_reasoning_drop_warning(caplog: pytest.LogCaptureFixture) -> None: + import httpx + + detail: Final = [ + {"type": "extra_forbidden", "loc": ["body", "reasoning_effort"], "msg": "Extra inputs are not permitted"} + ] + error: Final = httpx.HTTPStatusError( + message="422", + request=httpx.Request("POST", "https://example.services.ai.azure.com/models/chat/completions"), + response=httpx.Response(422, text=json.dumps({"error": {"message": json.dumps({"detail": detail})}})), + ) + request_data: Final = { + "model": "Mistral-Large-3", + "messages": [{"role": "user", "content": "Say hi."}], + "reasoning_effort": "high", + } + + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + retried: Final = AzureAIStudioConfig().transform_request_on_unprocessable_entity_error(error, request_data) + drop_warnings: Final = [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] + + assert retried == {"model": "Mistral-Large-3", "messages": [{"role": "user", "content": "Say hi."}]} + assert len(drop_warnings) == 1 + assert all( + fragment in drop_warnings[0] for fragment in ("reasoning_effort='high'", "Mistral-Large-3", "(drop_params=True)") + ) + + def _find_key_anywhere(obj, key: str) -> bool: if isinstance(obj, dict): if key in obj: diff --git a/tests/unit/llms/bedrock/chat/chat_completions/test_bedrock_chat_completions_transformation.py b/tests/unit/llms/bedrock/chat/chat_completions/test_bedrock_chat_completions_transformation.py index 3faae2e3289..f4dcff1969e 100644 --- a/tests/unit/llms/bedrock/chat/chat_completions/test_bedrock_chat_completions_transformation.py +++ b/tests/unit/llms/bedrock/chat/chat_completions/test_bedrock_chat_completions_transformation.py @@ -1,6 +1,7 @@ """Bedrock Runtime Chat Completions: the default for GPT 5.6 and newer, ``bedrock/chat_completions/`` for the rest.""" import json +import logging import re import httpx @@ -706,18 +707,30 @@ def test_with_max_completion_tokens_leaves_other_params_alone(): @pytest.mark.parametrize( - "model", - ["us.xai.grok-4.6", "bedrock/us-gov-west-1/us.xai.grok-4.6"], + ("model", "logged_model"), + [ + ("us.xai.grok-4.6", "us.xai.grok-4.6"), + ("chat_completions/us.xai.grok-4.6", "us.xai.grok-4.6"), + ("bedrock/us-gov-west-1/us.xai.grok-4.6", "us-gov-west-1/us.xai.grok-4.6"), + ], ) -def test_map_openai_params_drops_reasoning_effort_none_for_grok(model): +def test_map_openai_params_drops_reasoning_effort_none_for_grok_with_one_warning( + model: str, logged_model: str, caplog: pytest.LogCaptureFixture +) -> None: cfg = AmazonBedrockRuntimeChatCompletionsConfig() - mapped = cfg.map_openai_params( - non_default_params={"reasoning_effort": "none", "max_tokens": 64}, - optional_params={}, - model=model, - drop_params=False, - ) + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + mapped = cfg.map_openai_params( + non_default_params={"reasoning_effort": "none", "max_tokens": 64}, + optional_params={}, + model=model, + drop_params=False, + ) + drop_warnings = [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] assert "reasoning_effort" not in mapped + assert len(drop_warnings) == 1 + assert "reasoning_effort='none'" in drop_warnings[0] + assert f"model={logged_model!r}" in drop_warnings[0] + assert "drop_params" not in drop_warnings[0] def test_map_openai_params_keeps_reasoning_effort_low_for_grok(): @@ -750,17 +763,24 @@ def test_map_openai_params_refuses_a_non_string_reasoning_effort_without_drop_pa @pytest.mark.parametrize("reasoning_effort", [["low"], {"effort": "low"}, 5], ids=["list", "object", "int"]) @pytest.mark.parametrize("drop_params_via", ["request", "litellm.drop_params"]) def test_map_openai_params_drops_a_non_string_reasoning_effort_under_drop_params( - monkeypatch, model, reasoning_effort, drop_params_via + monkeypatch, caplog, model, reasoning_effort, drop_params_via ): monkeypatch.setattr(litellm, "drop_params", drop_params_via == "litellm.drop_params") - mapped = AmazonBedrockRuntimeChatCompletionsConfig().map_openai_params( - non_default_params={"reasoning_effort": reasoning_effort, "max_tokens": 64}, - optional_params={}, - model=model, - drop_params=drop_params_via == "request", - ) + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + mapped = AmazonBedrockRuntimeChatCompletionsConfig().map_openai_params( + non_default_params={"reasoning_effort": reasoning_effort, "max_tokens": 64}, + optional_params={}, + model=model, + drop_params=drop_params_via == "request", + ) + drop_warnings = [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] assert "reasoning_effort" not in mapped assert mapped["max_completion_tokens"] == 64 + assert len(drop_warnings) == 1 + assert all( + fragment in drop_warnings[0] + for fragment in (f"reasoning_effort={reasoning_effort!r}", repr(model), "(drop_params=True)") + ) def test_map_openai_params_keeps_reasoning_effort_none_for_gpt56(): diff --git a/tests/unit/llms/bedrock/chat/test_converse_transformation.py b/tests/unit/llms/bedrock/chat/test_converse_transformation.py index 743e7e74431..524caaba7a1 100644 --- a/tests/unit/llms/bedrock/chat/test_converse_transformation.py +++ b/tests/unit/llms/bedrock/chat/test_converse_transformation.py @@ -13,6 +13,7 @@ import pytest import respx import litellm +from litellm._logging import verbose_logger from litellm import ModelResponse, Router, completion from litellm.litellm_core_utils.prompt_templates.factory import ( _bedrock_converse_messages_pt, @@ -11096,6 +11097,52 @@ def test_a_deployment_with_a_batch_control_plane_endpoint_keeps_it_out_of_chat_r assert "aws_bedrock_control_plane_endpoint" not in route.calls[0].request.content.decode() +def _warnings_naming(caplog: pytest.LogCaptureFixture, param: str) -> tuple[str, ...]: + return tuple( + r.getMessage() for r in caplog.records if r.levelno == logging.WARNING and f"Dropping {param}=" in r.getMessage() + ) + + +_DEEPSEEK_R1_PROFILE_ARN: Final = "arn:aws:bedrock:us-east-1:123456789012:inference-profile/us.deepseek.r1-v1:0" + + +@pytest.mark.parametrize( + "param, value", (("thinking", {"type": "enabled", "budget_tokens": 1024}), ("reasoning_effort", "high")) +) +def test_bedrock_deepseek_arn_reasoning_drop_warns_once( + param: str, value: object, caplog: pytest.LogCaptureFixture +) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + optional_params: Final = AmazonConverseConfig().map_openai_params( + non_default_params={param: value, "max_tokens": 100}, + optional_params={}, + model=_DEEPSEEK_R1_PROFILE_ARN, + drop_params=False, + ) + + warnings: Final = _warnings_naming(caplog, param) + assert param not in optional_params + assert len(warnings) == 1 + assert _DEEPSEEK_R1_PROFILE_ARN in warnings[0] + + +@pytest.mark.parametrize( + "param, value", (("thinking", {"type": "enabled", "budget_tokens": 1024}), ("reasoning_effort", "high")) +) +def test_bedrock_deepseek_r1_drop_of_an_unadvertised_param_does_not_warn( + param: str, value: object, caplog: pytest.LogCaptureFixture +) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + AmazonConverseConfig().map_openai_params( + non_default_params={param: value, "max_tokens": 100}, + optional_params={}, + model="us.deepseek.r1-v1:0", + drop_params=False, + ) + + assert _warnings_naming(caplog, param) == () + + def _converse_bound_thinking_request( headers: dict[str, str] | None, body_betas: tuple[str, ...] = (), drop_params: bool = True ) -> dict[str, object]: diff --git a/tests/unit/llms/deepseek/chat/test_deepseek_chat_transformation.py b/tests/unit/llms/deepseek/chat/test_deepseek_chat_transformation.py index 3fbb12881da..56af47aeebe 100644 --- a/tests/unit/llms/deepseek/chat/test_deepseek_chat_transformation.py +++ b/tests/unit/llms/deepseek/chat/test_deepseek_chat_transformation.py @@ -1,4 +1,5 @@ import json +import logging from collections.abc import Iterator from typing import Final @@ -6,6 +7,7 @@ import httpx import litellm import pytest import respx +from litellm._logging import verbose_logger from litellm.caching.llm_caching_handler import LLMClientCache from litellm.llms.deepseek.chat.transformation import DeepSeekChatConfig @@ -951,3 +953,42 @@ def test_get_complete_url_ignores_a_trailing_slash_on_api_base(api_base: str): ) == "https://gateway.example.com/deepseek/chat/completions" ) + + +def _budget_warnings(caplog: pytest.LogCaptureFixture) -> tuple[str, ...]: + return tuple( + r.getMessage() + for r in caplog.records + if r.levelno == logging.WARNING and "Dropping thinking.budget_tokens=" in r.getMessage() + ) + + +def test_enabled_thinking_budget_dropped_for_deepseek_warns_once(caplog: pytest.LogCaptureFixture) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + result: Final = DeepSeekChatConfig().map_openai_params( + non_default_params={"thinking": {"type": "enabled", "budget_tokens": 4000}}, + optional_params={}, + model="deepseek-reasoner", + drop_params=False, + ) + + warnings: Final = _budget_warnings(caplog) + assert result["thinking"] == {"type": "enabled"} + assert len(warnings) == 1 + assert "budget_tokens=4000" in warnings[0] + assert "deepseek-reasoner" in warnings[0] + + +@pytest.mark.parametrize( + "thinking", ({"type": "enabled"}, {"type": "disabled", "budget_tokens": 4000}), ids=("no-budget", "disabled") +) +def test_deepseek_thinking_without_a_requested_budget_does_not_warn( + thinking: dict[str, object], caplog: pytest.LogCaptureFixture +) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + result: Final = DeepSeekChatConfig().map_openai_params( + non_default_params={"thinking": thinking}, optional_params={}, model="deepseek-reasoner", drop_params=False + ) + + assert result["thinking"] == {"type": thinking["type"]} + assert _budget_warnings(caplog) == () diff --git a/tests/unit/llms/fal_ai/chat/test_fal_ai_chat_transformation.py b/tests/unit/llms/fal_ai/chat/test_fal_ai_chat_transformation.py index b7a79f03f8d..f4a7455ff27 100644 --- a/tests/unit/llms/fal_ai/chat/test_fal_ai_chat_transformation.py +++ b/tests/unit/llms/fal_ai/chat/test_fal_ai_chat_transformation.py @@ -1,3 +1,6 @@ +import logging +from typing import Final + import httpx import pytest @@ -242,11 +245,33 @@ def test_map_openai_params_maps_reasoning_effort(effort, expected): assert mapped["reasoning"] is expected -def test_map_openai_params_drops_unknown_reasoning_effort_when_dropping(): - mapped = FalAIChatConfig().map_openai_params( - non_default_params={"reasoning_effort": "extreme"}, optional_params={}, model=MODEL, drop_params=True - ) +@pytest.mark.parametrize("drop_globally", [False, True]) +def test_map_openai_params_drops_unknown_reasoning_effort_with_one_warning( + drop_globally: bool, monkeypatch: pytest.MonkeyPatch, caplog: pytest.LogCaptureFixture +) -> None: + monkeypatch.setattr(litellm, "drop_params", drop_globally) + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + mapped: Final = FalAIChatConfig().map_openai_params( + non_default_params={"reasoning_effort": "xhigh"}, + optional_params={}, + model=MODEL, + drop_params=not drop_globally, + ) + drop_warnings: Final = [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] assert "reasoning" not in mapped + assert len(drop_warnings) == 1 + assert all(fragment in drop_warnings[0] for fragment in ("reasoning_effort='xhigh'", MODEL, "(drop_params=True)")) + + +def test_map_openai_params_rejects_unknown_reasoning_effort_without_drop_params( + monkeypatch: pytest.MonkeyPatch, +) -> None: + monkeypatch.setattr(litellm, "drop_params", False) + with pytest.raises(FalAIError) as exc_info: + FalAIChatConfig().map_openai_params( + non_default_params={"reasoning_effort": "xhigh"}, optional_params={}, model=MODEL, drop_params=False + ) + assert exc_info.value.status_code == 400 @pytest.mark.parametrize("effort", [{"level": "low"}, ["low"], 1]) diff --git a/tests/unit/llms/fireworks_ai/chat/test_fireworks_ai_chat_transformation.py b/tests/unit/llms/fireworks_ai/chat/test_fireworks_ai_chat_transformation.py index 28f4fdead9e..ed166d206c8 100644 --- a/tests/unit/llms/fireworks_ai/chat/test_fireworks_ai_chat_transformation.py +++ b/tests/unit/llms/fireworks_ai/chat/test_fireworks_ai_chat_transformation.py @@ -1,4 +1,5 @@ import json +import logging from collections.abc import Callable from typing import Final from unittest.mock import MagicMock, patch @@ -9,6 +10,7 @@ import respx from pydantic import ValidationError import litellm +from litellm._logging import verbose_logger from litellm.constants import SESSION_ID_GENERATED_METADATA_KEY from litellm.litellm_core_utils.get_supported_openai_params import get_supported_openai_params from litellm.llms.custom_httpx.http_handler import HTTPHandler @@ -2361,6 +2363,39 @@ def test_global_disable_flag_with_transform_messages_helper( assert json.loads(route.calls.last.request.content)["messages"] == [{"role": "user", "content": content}] +def _warnings_naming(caplog: pytest.LogCaptureFixture, param: str) -> tuple[str, ...]: + return tuple( + r.getMessage() for r in caplog.records if r.levelno == logging.WARNING and f"Dropping {param}=" in r.getMessage() + ) + + +def test_chat_template_kwargs_effort_dropped_for_non_reasoning_model_warns_once( + caplog: pytest.LogCaptureFixture, +) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + FireworksAIConfig().map_extra_body_params( + {"extra_body": {"chat_template_kwargs": {"reasoning_budget": 512}}}, + _NON_REASONING_MODEL, + ) + + warnings: Final = _warnings_naming(caplog, "chat_template_kwargs") + assert len(warnings) == 1 + assert "'reasoning_budget': 512" in warnings[0] + assert "llama-v3-70b-instruct" in warnings[0] + + +def test_chat_template_kwargs_thinking_off_for_non_reasoning_model_does_not_warn( + caplog: pytest.LogCaptureFixture, +) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + FireworksAIConfig().map_extra_body_params( + {"extra_body": {"chat_template_kwargs": {"enable_thinking": False}}}, + _NON_REASONING_MODEL, + ) + + assert _warnings_naming(caplog, "chat_template_kwargs") == () + + _CODEX_DEVELOPER_MESSAGE_AFTER_A_USER_TURN: Final = ( {"role": "system", "content": "You are Codex"}, {"role": "user", "content": "Hi there"}, diff --git a/tests/unit/llms/fireworks_ai/completion/test_fireworks_ai_text_completion_transformation.py b/tests/unit/llms/fireworks_ai/completion/test_fireworks_ai_text_completion_transformation.py index e5a77aa8d41..d35442d4095 100644 --- a/tests/unit/llms/fireworks_ai/completion/test_fireworks_ai_text_completion_transformation.py +++ b/tests/unit/llms/fireworks_ai/completion/test_fireworks_ai_text_completion_transformation.py @@ -1,7 +1,11 @@ +import logging +from typing import Final + import pytest import litellm +from litellm._logging import verbose_logger from litellm.llms.fireworks_ai.completion.transformation import ( @@ -204,3 +208,36 @@ def test_transform_text_completion_request_keeps_sdk_rejected_keys_in_extra_body assert "chat_template_kwargs" not in data["extra_body"] assert "best_of" not in data["extra_body"] assert "response_format" not in data + + +def _warnings_naming(caplog: pytest.LogCaptureFixture, param: str) -> tuple[str, ...]: + return tuple( + r.getMessage() for r in caplog.records if r.levelno == logging.WARNING and f"Dropping {param}=" in r.getMessage() + ) + + +def test_chat_template_kwargs_effort_dropped_for_non_reasoning_model_warns_once( + caplog: pytest.LogCaptureFixture, +) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + FireworksAITextCompletionConfig().map_extra_body_params( + {"extra_body": {"chat_template_kwargs": {"reasoning_budget": 512}}}, + _NON_REASONING_MODEL, + ) + + warnings: Final = _warnings_naming(caplog, "chat_template_kwargs") + assert len(warnings) == 1 + assert "'reasoning_budget': 512" in warnings[0] + assert "llama-v3-70b-instruct" in warnings[0] + + +def test_chat_template_kwargs_thinking_off_for_non_reasoning_model_does_not_warn( + caplog: pytest.LogCaptureFixture, +) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + FireworksAITextCompletionConfig().map_extra_body_params( + {"extra_body": {"chat_template_kwargs": {"enable_thinking": False}}}, + _NON_REASONING_MODEL, + ) + + assert _warnings_naming(caplog, "chat_template_kwargs") == () diff --git a/tests/unit/llms/mistral/test_mistral_chat_transformation.py b/tests/unit/llms/mistral/test_mistral_chat_transformation.py index c340f77c317..720a1b4f458 100644 --- a/tests/unit/llms/mistral/test_mistral_chat_transformation.py +++ b/tests/unit/llms/mistral/test_mistral_chat_transformation.py @@ -1,4 +1,5 @@ import json +import logging from collections.abc import Mapping from typing import Final, List, cast from unittest.mock import MagicMock, patch @@ -9,6 +10,7 @@ import pytest import respx from openai import AsyncOpenAI +from litellm._logging import verbose_logger from litellm.litellm_core_utils.prompt_templates.common_utils import TOOL_RESULT_IMAGE_BOUNDARY from litellm.llms.mistral.chat.transformation import ( MistralChatResponseIterator, @@ -1670,6 +1672,109 @@ def test_mistral_transform_request_drops_an_empty_tool_message_whose_call_id_is_ assert [m.get("role") for m in request["messages"]] == ["user", "assistant"] +def _warnings_naming(caplog: pytest.LogCaptureFixture, param: str) -> tuple[str, ...]: + return tuple( + r.getMessage() for r in caplog.records if r.levelno == logging.WARNING and f"Dropping {param}=" in r.getMessage() + ) + + +@pytest.mark.parametrize( + "param, value", + ( + ("reasoning_effort", "none"), + ("reasoning_effort", "xhigh"), + ("thinking", {"type": "disabled"}), + ("thinking", {"type": ["enabled"]}), + ), +) +def test_magistral_reasoning_request_replaced_by_prompt_warns_once( + param: str, value: object, caplog: pytest.LogCaptureFixture +) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + result: Final = MistralConfig().map_openai_params( + non_default_params={param: value}, + optional_params={}, + model="magistral-medium-latest", + drop_params=False, + ) + + warnings: Final = _warnings_naming(caplog, param) + assert result["_add_reasoning_prompt"] is True + assert len(warnings) == 1 + assert "magistral-medium-latest" in warnings[0] + + +@pytest.mark.parametrize( + "param, value", + ( + ("reasoning_effort", "low"), + ("reasoning_effort", "medium"), + ("reasoning_effort", "high"), + ("thinking", {"type": "enabled", "budget_tokens": 1024}), + ), +) +def test_magistral_reasoning_request_the_prompt_honors_does_not_warn( + param: str, value: object, caplog: pytest.LogCaptureFixture +) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + result: Final = MistralConfig().map_openai_params( + non_default_params={param: value}, + optional_params={}, + model="magistral-medium-latest", + drop_params=False, + ) + + assert result["_add_reasoning_prompt"] is True + assert _warnings_naming(caplog, param) == () + + +def test_non_magistral_reasoning_effort_is_sent_without_warning(caplog: pytest.LogCaptureFixture) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + result: Final = MistralConfig().map_openai_params( + non_default_params={"reasoning_effort": "high"}, + optional_params={}, + model="mistral-medium-latest", + drop_params=False, + ) + + assert result["reasoning_effort"] == "high" + assert _warnings_naming(caplog, "reasoning_effort") == () + + +@pytest.mark.parametrize("reasoning_effort", [5, {"effort": 5}, {"effort": 5, "summary": "auto"}]) +def test_mistral_warns_once_when_drop_params_drops_a_level_it_cannot_send( + caplog: pytest.LogCaptureFixture, reasoning_effort: object +) -> None: + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + result: Final = MistralConfig().map_openai_params( + non_default_params={"reasoning_effort": reasoning_effort}, + optional_params={}, + model="mistral-medium-latest", + drop_params=True, + ) + + assert "reasoning_effort" not in result + warnings: Final = _warnings_naming(caplog, "reasoning_effort") + assert len(warnings) == 1 + assert "model='mistral-medium-latest'" in warnings[0] + assert "(drop_params=True)" in warnings[0] + + +@pytest.mark.parametrize("reasoning_effort", [{"summary": "auto"}, {"effort": "high", "summary": "auto"}]) +def test_mistral_stays_quiet_under_drop_params_when_no_level_is_lost( + caplog: pytest.LogCaptureFixture, reasoning_effort: object +) -> None: + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + MistralConfig().map_openai_params( + non_default_params={"reasoning_effort": reasoning_effort}, + optional_params={}, + model="mistral-medium-latest", + drop_params=True, + ) + + assert _warnings_naming(caplog, "reasoning_effort") == () + + @pytest.mark.parametrize( ("model", "custom_llm_provider"), [ diff --git a/tests/unit/llms/oci/chat/test_oci_chat_transformation.py b/tests/unit/llms/oci/chat/test_oci_chat_transformation.py index 0535279d177..f22b5e32593 100644 --- a/tests/unit/llms/oci/chat/test_oci_chat_transformation.py +++ b/tests/unit/llms/oci/chat/test_oci_chat_transformation.py @@ -1,4 +1,5 @@ import datetime +import logging from typing import Final import httpx import pytest @@ -1103,19 +1104,25 @@ class TestOCIReasoningEffort: ) assert chat_request.get("reasoningEffort") == "HIGH" - def test_reasoning_effort_unsupported_on_cohere_dropped(self): - """drop_params=True → silently drop reasoning_effort for Cohere.""" + def test_reasoning_effort_unsupported_on_cohere_dropped_with_one_warning(self, caplog): from litellm.llms.oci.chat.transformation import OCIChatConfig config = OCIChatConfig() - result = config.map_openai_params( - non_default_params={"reasoning_effort": "low"}, - optional_params={}, - model="cohere.command-latest", - drop_params=True, - ) + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + result = config.map_openai_params( + non_default_params={"reasoning_effort": "low"}, + optional_params={}, + model="cohere.command-latest", + drop_params=True, + ) + drop_warnings = [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] assert "reasoning_effort" not in result assert "reasoningEffort" not in result + assert len(drop_warnings) == 1 + assert all( + fragment in drop_warnings[0] + for fragment in ("reasoning_effort='low'", "cohere.command-latest", "(drop_params=True)") + ) def test_reasoning_effort_unsupported_on_cohere_raises(self): """drop_params=False → raise rather than ship a payload Cohere will reject.""" @@ -2110,16 +2117,20 @@ class TestReasoningEffortGating: assert "reasoning_effort" not in OCIChatConfig().get_supported_openai_params("meta.llama-3.3-70b-instruct") @pytest.mark.parametrize("model", SERVED_GROK_4_20_MODELS) - def test_reasoning_effort_dropped_under_drop_params(self, local_model_cost_map, model): - result = OCIChatConfig().map_openai_params( - non_default_params={"reasoning_effort": "low", "temperature": 0.1}, - optional_params={}, - model=model, - drop_params=True, - ) + def test_reasoning_effort_dropped_under_drop_params_with_one_warning(self, local_model_cost_map, model, caplog): + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + result = OCIChatConfig().map_openai_params( + non_default_params={"reasoning_effort": "low", "temperature": 0.1}, + optional_params={}, + model=model, + drop_params=True, + ) + drop_warnings = [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] assert "reasoningEffort" not in result assert "reasoning_effort" not in result assert result["temperature"] == 0.1 + assert len(drop_warnings) == 1 + assert all(fragment in drop_warnings[0] for fragment in ("reasoning_effort='low'", model, "(drop_params=True)")) def test_reasoning_effort_raises_without_drop_params_for_gated_grok(self, local_model_cost_map, monkeypatch): monkeypatch.setattr(litellm, "drop_params", False) diff --git a/tests/unit/llms/ollama/test_ollama_chat_transformation.py b/tests/unit/llms/ollama/test_ollama_chat_transformation.py index df9f5fd9d14..af1b85cf325 100644 --- a/tests/unit/llms/ollama/test_ollama_chat_transformation.py +++ b/tests/unit/llms/ollama/test_ollama_chat_transformation.py @@ -1,5 +1,6 @@ import asyncio import inspect +import logging import os import sys from typing import Final, cast @@ -23,6 +24,7 @@ import json from unittest.mock import ANY, MagicMock, patch import litellm +from litellm._logging import verbose_logger from litellm.types.utils import Choices, EmbeddingResponse, Message, ModelResponse, ModelResponseStream from unittest import mock @@ -1114,3 +1116,36 @@ def test_get_complete_url_ignores_a_trailing_slash_on_api_base(api_base: str): ) == "http://gateway.example.com/ollama/api/chat" ) + + +def _warnings_naming(caplog: pytest.LogCaptureFixture, param: str) -> tuple[str, ...]: + return tuple( + r.getMessage() for r in caplog.records if r.levelno == logging.WARNING and f"Dropping {param}=" in r.getMessage() + ) + + +@pytest.mark.parametrize("effort", ("minimal", "xhigh")) +def test_reasoning_effort_mapped_to_think_off_warns_once(effort: str, caplog: pytest.LogCaptureFixture) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + result: Final = OllamaChatConfig().map_openai_params( + non_default_params={"reasoning_effort": effort}, optional_params={}, model="qwen3", drop_params=False + ) + + warnings: Final = _warnings_naming(caplog, "reasoning_effort") + assert result["think"] is False + assert len(warnings) == 1 + assert f"reasoning_effort={effort!r}" in warnings[0] + assert "qwen3" in warnings[0] + + +@pytest.mark.parametrize("effort, think", (("none", False), ("low", True), ("high", True))) +def test_reasoning_effort_with_a_matching_think_value_does_not_warn( + effort: str, think: bool, caplog: pytest.LogCaptureFixture +) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + result: Final = OllamaChatConfig().map_openai_params( + non_default_params={"reasoning_effort": effort}, optional_params={}, model="qwen3", drop_params=False + ) + + assert result["think"] is think + assert _warnings_naming(caplog, "reasoning_effort") == () diff --git a/tests/unit/llms/ollama/test_ollama_completion_transformation.py b/tests/unit/llms/ollama/test_ollama_completion_transformation.py index 248e745ebff..edea39bb10c 100644 --- a/tests/unit/llms/ollama/test_ollama_completion_transformation.py +++ b/tests/unit/llms/ollama/test_ollama_completion_transformation.py @@ -1,6 +1,7 @@ import base64 import io import json +import logging import sys from typing import Final from unittest.mock import MagicMock, patch @@ -10,6 +11,7 @@ import pytest import respx import litellm +from litellm._logging import verbose_logger from litellm._uuid import uuid from litellm.llms.custom_httpx.http_handler import AsyncHTTPHandler from litellm.llms.ollama.completion.transformation import ( @@ -990,3 +992,40 @@ def test_get_complete_url_ignores_a_trailing_slash_on_api_base(api_base: str): ) == "http://gateway.example.com/ollama/api/generate" ) + +def _reasoning_effort_warnings(caplog: pytest.LogCaptureFixture) -> tuple[str, ...]: + return tuple( + r.getMessage() + for r in caplog.records + if r.levelno == logging.WARNING and "Dropping reasoning_effort=" in r.getMessage() + ) + + +@pytest.mark.parametrize("effort", ("minimal", "xhigh")) +def test_reasoning_effort_mapped_to_think_off_warns_once(effort: str, caplog: pytest.LogCaptureFixture) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + result: Final = OllamaConfig().map_openai_params( + non_default_params={"reasoning_effort": effort}, optional_params={}, model="qwen3:0.6b", drop_params=False + ) + + warnings: Final = _reasoning_effort_warnings(caplog) + assert result["think"] is False + assert len(warnings) == 1 + assert f"reasoning_effort={effort!r}" in warnings[0] + assert "qwen3:0.6b" in warnings[0] + + +@pytest.mark.parametrize( + "model, effort, think", + (("qwen3:0.6b", "none", False), ("qwen3:0.6b", "high", True), ("gpt-oss:20b", "xhigh", "xhigh")), +) +def test_reasoning_effort_ollama_honors_does_not_warn( + model: str, effort: str, think: object, caplog: pytest.LogCaptureFixture +) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + result: Final = OllamaConfig().map_openai_params( + non_default_params={"reasoning_effort": effort}, optional_params={}, model=model, drop_params=False + ) + + assert result["think"] == think + assert _reasoning_effort_warnings(caplog) == () diff --git a/tests/unit/llms/openai/test_openai_common_utils.py b/tests/unit/llms/openai/test_openai_common_utils.py index 46c14fcfd8f..f0865cca7f9 100644 --- a/tests/unit/llms/openai/test_openai_common_utils.py +++ b/tests/unit/llms/openai/test_openai_common_utils.py @@ -1,4 +1,6 @@ import asyncio +import json +import logging from typing import Final from unittest.mock import MagicMock, patch @@ -9,7 +11,11 @@ import respx import litellm from litellm.litellm_core_utils.token_counter import token_counter -from litellm.llms.openai.common_utils import BaseOpenAILLM, is_openai_backed_api_base +from litellm.llms.openai.common_utils import ( + BaseOpenAILLM, + drop_params_from_unprocessable_entity_error, + is_openai_backed_api_base, +) # Test parameters for different API functions API_FUNCTION_PARAMS = [ @@ -319,6 +325,104 @@ def _completion_kwargs(provider: str, client, **overrides) -> dict: } +_REASONING_EFFORT_422: Final = { + "error": { + "message": json.dumps( + {"detail": [{"loc": ["body", "reasoning_effort"], "msg": "Extra inputs are not permitted"}]} + ) + } +} +_CHAT_COMPLETION: Final = { + "id": "chatcmpl-1", + "object": "chat.completion", + "created": 1, + "model": "gpt-6.1-sol", + "choices": [{"index": 0, "message": {"role": "assistant", "content": "ok"}, "finish_reason": "stop"}], + "usage": {"prompt_tokens": 1, "completion_tokens": 1, "total_tokens": 2}, +} +_CHAT_COMPLETION_STREAM: Final = ( + 'data: {"id":"chatcmpl-1","object":"chat.completion.chunk","created":1,"model":"gpt-6.1-sol",' + '"choices":[{"index":0,"delta":{"role":"assistant","content":"ok"},"finish_reason":"stop"}]}\n\n' + "data: [DONE]\n\n" +) + + +def _answer_422_while_reasoning_effort_is_sent(request: httpx.Request) -> httpx.Response: + body: Final = json.loads(request.content) + if "reasoning_effort" in body: + return httpx.Response(422, json=_REASONING_EFFORT_422) + if body.get("stream"): + return httpx.Response(200, text=_CHAT_COMPLETION_STREAM, headers={"content-type": "text/event-stream"}) + return httpx.Response(200, json=_CHAT_COMPLETION) + + +async def _complete_through_the_422_retry(call_shape: str) -> str: + transport: Final = httpx.MockTransport(_answer_422_while_reasoning_effort_is_sent) + kwargs: Final = { + "model": "gpt-6.1-sol", + "messages": [{"role": "user", "content": "hi"}], + "reasoning_effort": "high", + } + if call_shape == "sync": + sync_client: Final = openai.OpenAI(api_key="sk-not-a-real-key", http_client=httpx.Client(transport=transport)) + return litellm.completion(**kwargs, client=sync_client).choices[0].message.content + async_client: Final = openai.AsyncOpenAI( + api_key="sk-not-a-real-key", http_client=httpx.AsyncClient(transport=transport) + ) + if call_shape == "async": + return (await litellm.acompletion(**kwargs, client=async_client)).choices[0].message.content + stream: Final = await litellm.acompletion(**kwargs, client=async_client, stream=True) + return "".join([chunk.choices[0].delta.content or "" async for chunk in stream]) + + +@pytest.mark.parametrize("call_shape", ["sync", "async", "async_stream"]) +def test_a_422_naming_reasoning_effort_drops_it_with_one_warning( + call_shape: str, monkeypatch: pytest.MonkeyPatch, caplog: pytest.LogCaptureFixture +) -> None: + monkeypatch.setattr(litellm, "drop_params", True) + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + content: Final = asyncio.run(_complete_through_the_422_retry(call_shape)) + drop_warnings: Final = [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] + + assert content == "ok" + assert len(drop_warnings) == 1 + assert all( + fragment in drop_warnings[0] + for fragment in ("reasoning_effort='high'", "gpt-6.1-sol", "(drop_params=True)", "422") + ) + + +def test_a_422_with_an_unhashable_loc_still_drops_reasoning_effort_with_one_warning( + monkeypatch: pytest.MonkeyPatch, caplog: pytest.LogCaptureFixture +) -> None: + detail: Final = [ + {"loc": ["body", ["messages", 0]], "msg": "Invalid"}, + {"loc": ["body", "reasoning_effort"], "msg": "Extra inputs are not permitted"}, + ] + + def answer(request: httpx.Request) -> httpx.Response: + if "reasoning_effort" in json.loads(request.content): + return httpx.Response(422, json={"error": {"message": json.dumps({"detail": detail})}}) + return httpx.Response(200, json=_CHAT_COMPLETION) + + monkeypatch.setattr(litellm, "drop_params", True) + client: Final = openai.OpenAI( + api_key="sk-not-a-real-key", http_client=httpx.Client(transport=httpx.MockTransport(answer)) + ) + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + response: Final = litellm.completion( + model="gpt-6.1-sol", + messages=[{"role": "user", "content": "hi"}], + reasoning_effort="high", + client=client, + ) + drop_warnings: Final = [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] + + assert response.choices[0].message.content == "ok" + assert len(drop_warnings) == 1 + assert "reasoning_effort='high'" in drop_warnings[0] + + @pytest.mark.parametrize("provider", ["openai", "azure"]) def test_sync_output_limit_400_maps_to_length_truncated_response(provider): response = litellm.completion( @@ -588,3 +692,24 @@ async def test_azure_gateway_and_image_variations_use_the_callers_async_session( ) assert response.data[0].url == "https://example.com/image.png" assert not session.is_closed + + + +def test_reasoning_drop_warning_caps_a_large_caller_value(caplog: pytest.LogCaptureFixture) -> None: + error: Final = httpx.HTTPStatusError( + message="422", + request=httpx.Request("POST", "https://example.invalid/v1/chat/completions"), + response=httpx.Response(422, json=_REASONING_EFFORT_422), + ) + data: Final[dict[str, object]] = {"model": "gpt-6.1-sol", "messages": [], "reasoning_effort": "x" * 5000} + + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + retried: Final = drop_params_from_unprocessable_entity_error( + error, data, model="gpt-6.1-sol", drop_params_gated=True + ) + drop_warnings: Final = [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] + + assert "reasoning_effort" not in retried + assert len(drop_warnings) == 1 + assert len(drop_warnings[0]) < 300 + assert all(fragment in drop_warnings[0] for fragment in ("reasoning_effort='xxx", "'gpt-6.1-sol'", "422")) diff --git a/tests/unit/llms/tencent/chat/test_tencent_chat_transformation.py b/tests/unit/llms/tencent/chat/test_tencent_chat_transformation.py index 4d6d252ae6e..1bf646fc2ad 100644 --- a/tests/unit/llms/tencent/chat/test_tencent_chat_transformation.py +++ b/tests/unit/llms/tencent/chat/test_tencent_chat_transformation.py @@ -1,5 +1,10 @@ +import logging +from typing import Final from unittest.mock import patch +import pytest + +from litellm._logging import verbose_logger from litellm.llms.tencent.chat.transformation import TencentChatConfig @@ -208,6 +213,27 @@ class TestAdaptiveThinkingCoercion: assert result["extra_body"]["thinking"] == {"type": "adaptive", "budget_tokens": 4096} + @pytest.mark.parametrize("budget, warned", (("high", 1), (4096, 0)), ids=("non-integer", "integer")) + def test_enabled_thinking_on_adaptive_only_model_warns_only_when_the_budget_is_dropped( + self, local_model_cost_map, budget: object, warned: int, caplog: pytest.LogCaptureFixture + ): + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + result: Final = TencentChatConfig().map_openai_params( + non_default_params={"thinking": {"type": "enabled", "budget_tokens": budget}}, + optional_params={}, + model="tencent/minimax-m3", + drop_params=False, + ) + + warnings: Final = tuple( + r.getMessage() + for r in caplog.records + if r.levelno == logging.WARNING and f"Dropping thinking.budget_tokens={budget!r}" in r.getMessage() + ) + assert result["extra_body"]["thinking"]["type"] == "adaptive" + assert result["extra_body"]["thinking"].get("budget_tokens") == (None if warned else budget) + assert len(warnings) == warned + def test_disabled_thinking_kept_for_adaptive_only_model(self, local_model_cost_map): config = TencentChatConfig() result = config.map_openai_params( diff --git a/tests/unit/llms/vertex_ai/gemini/test_vertex_and_google_ai_studio_gemini.py b/tests/unit/llms/vertex_ai/gemini/test_vertex_and_google_ai_studio_gemini.py index 67c78b62151..98d1b8f1300 100644 --- a/tests/unit/llms/vertex_ai/gemini/test_vertex_and_google_ai_studio_gemini.py +++ b/tests/unit/llms/vertex_ai/gemini/test_vertex_and_google_ai_studio_gemini.py @@ -1693,6 +1693,78 @@ def test_vertex_ai_map_thinking_param_without_budget_tokens_for_gemini_3(): assert result["thinkingConfig"] == {"includeThoughts": True} +def _budget_drop_warnings(caplog: pytest.LogCaptureFixture) -> tuple[str, ...]: + return tuple( + r.getMessage() + for r in caplog.records + if r.levelno == logging.WARNING and "thinking.budget_tokens" in r.getMessage() + ) + + +@pytest.mark.parametrize("budget", ("high", 1024.5, {"tokens": 1024})) +def test_gemini_non_integer_thinking_budget_drop_warns_once(budget: object, caplog: pytest.LogCaptureFixture) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + result: Final = VertexGeminiConfig().map_openai_params( + non_default_params={"thinking": {"type": "enabled", "budget_tokens": budget}}, + optional_params={}, + model="gemini-2.5-flash", + drop_params=False, + ) + + warnings: Final = _budget_drop_warnings(caplog) + assert "thinkingBudget" not in result["thinkingConfig"] + assert len(warnings) == 1 + assert f"thinking.budget_tokens={budget!r}" in warnings[0] + assert "gemini-2.5-flash" in warnings[0] + + +def test_gemini_integer_thinking_budget_is_sent_without_warning(caplog: pytest.LogCaptureFixture) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + result: Final = VertexGeminiConfig().map_openai_params( + non_default_params={"thinking": {"type": "enabled", "budget_tokens": 2048}}, + optional_params={}, + model="gemini-2.5-flash", + drop_params=False, + ) + + assert result["thinkingConfig"]["thinkingBudget"] == 2048 + assert _budget_drop_warnings(caplog) == () + + +def test_gemini_3_thinking_budget_drop_warns_once(caplog: pytest.LogCaptureFixture) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + result: Final = VertexGeminiConfig().map_openai_params( + non_default_params={"thinking": {"type": "enabled", "budget_tokens": 2048}}, + optional_params={}, + model="gemini-3-pro-preview", + drop_params=False, + ) + + warnings: Final = _budget_drop_warnings(caplog) + assert "thinkingBudget" not in result["thinkingConfig"] + assert len(warnings) == 1 + assert "thinking.budget_tokens=2048" in warnings[0] + assert "gemini-3-pro-preview" in warnings[0] + + +@pytest.mark.parametrize( + "thinking", + ({"type": "enabled"}, {"type": "enabled", "budget_tokens": 0}, {"type": "disabled", "budget_tokens": 2048}), +) +def test_gemini_3_thinking_without_a_dropped_budget_does_not_warn( + thinking: dict[str, object], caplog: pytest.LogCaptureFixture +) -> None: + with caplog.at_level(logging.WARNING, logger=verbose_logger.name): + VertexGeminiConfig().map_openai_params( + non_default_params={"thinking": thinking}, + optional_params={}, + model="gemini-3-pro-preview", + drop_params=False, + ) + + assert _budget_drop_warnings(caplog) == () + + def test_vertex_ai_map_tools(): v = VertexGeminiConfig() optional_params = {} diff --git a/tests/unit/test_main.py b/tests/unit/test_main.py index aaec486b60f..9f2d2a9bbc5 100644 --- a/tests/unit/test_main.py +++ b/tests/unit/test_main.py @@ -10262,9 +10262,10 @@ def test_a_non_numeric_thinking_budget_on_adaptive_claude_is_rejected_with_a_400 @pytest.mark.parametrize("drop_globally", [False, True]) @pytest.mark.parametrize("model_info", [None, _OPUS_DECLARING_LOW_AND_HIGH]) @pytest.mark.parametrize("body_sent", [_anthropic_chat_body_sent, _bedrock_converse_request_fields_sent]) -def test_drop_params_drops_a_non_numeric_thinking_budget_on_adaptive_claude( +def test_drop_params_drops_a_non_numeric_thinking_budget_on_adaptive_claude_with_one_warning( respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch, + caplog: pytest.LogCaptureFixture, body_sent: Callable[..., Mapping[str, object]], model_info: Mapping[str, object] | None, drop_globally: bool, @@ -10272,13 +10273,40 @@ def test_drop_params_drops_a_non_numeric_thinking_budget_on_adaptive_claude( monkeypatch.setattr(litellm, "drop_params", drop_globally) request_drop: Final = {} if drop_globally else {"drop_params": True} - dropped: Final = body_sent( - respx_mock, model_info, thinking={"type": "enabled", "budget_tokens": "lots"}, **request_drop - ) + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + dropped: Final = body_sent( + respx_mock, model_info, thinking={"type": "enabled", "budget_tokens": "lots"}, **request_drop + ) + drop_warnings: Final = [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] adaptive_alone: Final = body_sent(respx_mock, model_info, thinking={"type": "adaptive"}, **request_drop) assert "output_config" not in adaptive_alone assert dropped == adaptive_alone + assert len(drop_warnings) == 1 + assert all( + fragment in drop_warnings[0] + for fragment in ("thinking.budget_tokens='lots'", "claude-opus-5-5", "(drop_params=True)") + ) + + +@pytest.mark.parametrize("model_info", [None, _KIMI_DEPLOYMENT]) +def test_drop_params_drops_a_non_numeric_thinking_budget_on_hosted_vllm_with_one_warning( + respx_mock: respx.MockRouter, caplog: pytest.LogCaptureFixture, model_info: Mapping[str, object] | None +) -> None: + with caplog.at_level(logging.WARNING, logger="LiteLLM"): + sent: Final = _chat_body_sent( + respx_mock, + "hosted_vllm/my-kimi", + None, + model_info, + thinking={"type": "enabled", "budget_tokens": "lots"}, + drop_params=True, + ) + + assert not {"reasoning_effort", "thinking"} & sent.keys() + drop_warnings: Final = [record.getMessage() for record in caplog.records if "Dropping" in record.getMessage()] + assert len(drop_warnings) == 1 + assert all(fragment in drop_warnings[0] for fragment in ("thinking.budget_tokens='lots'", "my-kimi")) @pytest.mark.parametrize("model", ["azure_ai/my-kimi", "hosted_vllm/my-kimi"]) From d18efdfe1362194edfb98d814c85307af35d2038 Mon Sep 17 00:00:00 2001 From: alvcol Date: Fri, 30 Oct 2026 16:57:48 +0100 Subject: [PATCH 08/18] fix(vllm): send the pass-through key as a bearer on both /vllm routes (#46861) * fix(vllm): send the deployment api_key on /vllm passthrough calls VLLMModelInfo.get_api_key always returned None, so /vllm/{endpoint} calls to a router model reached the vLLM server with no auth header even when the deployment had an api_key, and a server started with --api-key answered 401 (#23443). Return the deployment key (or VLLM_API_KEY) and send it as Authorization: Bearer, the header vllm serve --api-key checks. x-api-key was never actually sent before, since the key was always None. * fix(vllm): send the pass-through key as a bearer on both /vllm routes vllm serve --api-key only reads Authorization: Bearer, so a keyed vLLM answered 401 on both pass-through routes. Send the key as a bearer and keep x-api-key for gateways that read it. Router models send their own deployment key and never fall back to VLLM_API_KEY, which belongs to the /vllm route's own server. * fix(vllm): send only the bearer on router models and stop doubling /v1 on /vllm Router models now send the same Authorization: Bearer header /v1/chat/completions sends for that deployment, without the x-api-key copy that route never sent. The VLLM_API_BASE route keeps x-api-key next to the bearer, as it sent it before. A router model whose api_base already ends in /v1 no longer gets /v1/v1 on /vllm/v1 paths. * fix(vllm): send no placeholder key from keyless router models and stop doubling /v1 on the VLLM_API_BASE route * fix(vllm): send VLLM_API_KEY when a router model with no api_base is forwarded to VLLM_API_BASE * fix(vllm): keep a router model's own key on the VLLM_API_BASE server and resolve an oidc VLLM_API_KEY * fix(vllm): treat an empty key as no key so pass-through sends no empty bearer * fix(vllm): send no bearer for a Google oidc key and match VLLM_API_BASE across a trailing slash --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> --- litellm/llms/vllm/common_utils.py | 20 +- .../llms/vllm/passthrough/transformation.py | 35 +++- .../llm_passthrough_endpoints.py | 3 +- tests/unit/passthrough/test_main.py | 175 ++++++++++++++++++ .../test_llm_pass_through_endpoints.py | 62 +++++++ 5 files changed, 289 insertions(+), 6 deletions(-) diff --git a/litellm/llms/vllm/common_utils.py b/litellm/llms/vllm/common_utils.py index 6484cdc88e3..014fabd5187 100644 --- a/litellm/llms/vllm/common_utils.py +++ b/litellm/llms/vllm/common_utils.py @@ -1,3 +1,4 @@ +from collections.abc import Mapping from typing import Final import httpx @@ -10,6 +11,14 @@ from litellm.types.llms.openai import AllMessageValues from litellm.utils import add_path_to_api_base +def endpoint_under_api_base(endpoint: str, api_base: str | None) -> str: + return ( + endpoint.lstrip("/").removeprefix("v1/") + if api_base is not None and api_base.rstrip("/").endswith("/v1") + else endpoint + ) + + class VLLMError(BaseLLMException): def __init__( self, @@ -39,9 +48,14 @@ class VLLMModelInfo(BaseLLMModelInfo): api_key: str | None = None, api_base: str | None = None, ) -> dict: - if api_key is not None: - headers["x-api-key"] = api_key - return headers + upstream_key: Final = self.upstream_api_key(api_key, api_base) + return {**headers, **self.upstream_auth_headers(upstream_key)} if upstream_key else headers + + def upstream_api_key(self, api_key: str | None, api_base: str | None) -> str | None: + return api_key + + def upstream_auth_headers(self, api_key: str) -> Mapping[str, str]: + return {"Authorization": f"Bearer {api_key}", "x-api-key": api_key} @staticmethod def get_api_base(api_base: str | None = None) -> str | None: diff --git a/litellm/llms/vllm/passthrough/transformation.py b/litellm/llms/vllm/passthrough/transformation.py index e7da6e54410..2b253e7fa38 100644 --- a/litellm/llms/vllm/passthrough/transformation.py +++ b/litellm/llms/vllm/passthrough/transformation.py @@ -3,6 +3,7 @@ from typing import TYPE_CHECKING, Final from httpx import Response +from litellm.constants import PLACEHOLDER_API_KEY from litellm.litellm_core_utils.litellm_logging import Logging from litellm.llms.base_llm.passthrough.transformation import BasePassthroughConfig from litellm.llms.openai_like.passthrough.transformation import ( @@ -12,8 +13,14 @@ from litellm.llms.openai_like.passthrough.transformation import ( logged_text_completion_stream, relays_text_completion, ) +from litellm.secret_managers.main import ( + GOOGLE_OIDC_SECRET_PREFIX, + UnresolvableSecretReference, + get_secret_str, + resolve_local_oidc_reference, +) -from ..common_utils import VLLMModelInfo +from ..common_utils import VLLMModelInfo, endpoint_under_api_base if TYPE_CHECKING: from httpx import URL @@ -22,6 +29,30 @@ if TYPE_CHECKING: class VLLMPassthroughConfig(VLLMModelInfo, BasePassthroughConfig): + @staticmethod + def get_api_key(api_key: str | None = None) -> str | None: + return None if api_key == PLACEHOLDER_API_KEY else api_key + + def upstream_api_key(self, api_key: str | None, api_base: str | None) -> str | None: + if api_key is not None or not self._is_vllm_api_base(api_base): + return api_key + vllm_api_key: Final = get_secret_str("VLLM_API_KEY") + if vllm_api_key is None or vllm_api_key.startswith(GOOGLE_OIDC_SECRET_PREFIX): + return None + match resolve_local_oidc_reference(vllm_api_key): + case UnresolvableSecretReference(): + return None + case str() as resolved: + return resolved + + @staticmethod + def _is_vllm_api_base(api_base: str | None) -> bool: + vllm_api_base: Final = get_secret_str("VLLM_API_BASE") + return api_base is not None and vllm_api_base is not None and api_base.rstrip("/") == vllm_api_base.rstrip("/") + + def upstream_auth_headers(self, api_key: str) -> Mapping[str, str]: + return {"Authorization": f"Bearer {api_key}"} + def is_streaming_request(self, endpoint: str, request_data: dict) -> bool: return "stream" in request_data @@ -40,7 +71,7 @@ class VLLMPassthroughConfig(VLLMModelInfo, BasePassthroughConfig): raise Exception("VLLM api base not found") return ( - self.format_url(endpoint, base_target_url, request_query_params), + self.format_url(endpoint_under_api_base(endpoint, base_target_url), base_target_url, request_query_params), base_target_url, ) diff --git a/litellm/proxy/pass_through_endpoints/llm_passthrough_endpoints.py b/litellm/proxy/pass_through_endpoints/llm_passthrough_endpoints.py index 21078dfc848..8de7441121f 100644 --- a/litellm/proxy/pass_through_endpoints/llm_passthrough_endpoints.py +++ b/litellm/proxy/pass_through_endpoints/llm_passthrough_endpoints.py @@ -72,6 +72,7 @@ from litellm.llms.openai.common_utils import is_openai_agents_api_path, openai_a from litellm.llms.openai.workload_identity import get_workload_identity_bearer_token_for_api_base from litellm.llms.oss_decision import OssDecisionProvider, oss_connection, validate_oss_request from litellm.llms.vertex_ai.vertex_llm_base import VertexBase +from litellm.llms.vllm.common_utils import endpoint_under_api_base from litellm.passthrough.main import AsyncPassthroughStreamingResponse, is_provider_error from litellm.proxy._types import * from litellm.proxy.auth.auth_checks import enforced_model_allowlists @@ -600,7 +601,7 @@ async def vllm_proxy_route( ) return await llm_passthrough_factory_proxy_route( - endpoint=endpoint, + endpoint=endpoint_under_api_base(endpoint, get_secret_str("VLLM_API_BASE")), request=request, fastapi_response=fastapi_response, user_api_key_dict=user_api_key_dict, diff --git a/tests/unit/passthrough/test_main.py b/tests/unit/passthrough/test_main.py index 78fa83d126b..4997d736d0a 100644 --- a/tests/unit/passthrough/test_main.py +++ b/tests/unit/passthrough/test_main.py @@ -51,3 +51,178 @@ async def test_allm_passthrough_route_hosted_vllm_sends_normalized_model(respx_m outbound: Final = json.loads(route.calls[0].request.content) assert outbound["model"] == "my-deployment" assert outbound["messages"] == [{"role": "user", "content": "Hello"}] + + +@pytest.mark.asyncio +async def test_allm_passthrough_route_hosted_vllm_sends_deployment_api_key(respx_mock: respx.MockRouter) -> None: + route: Final = respx_mock.post(f"{_API_BASE}/v1/chat/completions").mock( + return_value=httpx.Response(200, json={"ok": True}) + ) + response: Final = await allm_passthrough_route( + method="POST", + endpoint="v1/chat/completions", + model="hosted_vllm/my-deployment", + api_base=_API_BASE, + api_key="sk-upstream", + json={"model": "anything", "messages": [{"role": "user", "content": "Hello"}]}, + client=AsyncHTTPHandler(), + ) + assert response.status_code == 200 + assert route.calls[0].request.headers["authorization"] == "Bearer sk-upstream" + assert "x-api-key" not in route.calls[0].request.headers + + +@pytest.mark.parametrize("endpoint", ["v1/chat/completions", "chat/completions"]) +@pytest.mark.asyncio +async def test_allm_passthrough_route_hosted_vllm_reaches_v1_paths_under_an_api_base_ending_in_v1( + respx_mock: respx.MockRouter, endpoint: str +) -> None: + route: Final = respx_mock.post(f"{_API_BASE}/v1/chat/completions").mock( + return_value=httpx.Response(200, json={"ok": True}) + ) + response: Final = await allm_passthrough_route( + method="POST", + endpoint=endpoint, + model="hosted_vllm/my-deployment", + api_base=f"{_API_BASE}/v1", + api_key="sk-upstream", + json={"model": "anything", "messages": [{"role": "user", "content": "Hello"}]}, + client=AsyncHTTPHandler(), + ) + assert response.status_code == 200 + assert route.call_count == 1 + + +@pytest.mark.asyncio +async def test_allm_passthrough_route_hosted_vllm_sends_no_auth_for_a_keyless_deployment( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch +) -> None: + monkeypatch.setenv("VLLM_API_KEY", "sk-vllm-route-only") + monkeypatch.delenv("HOSTED_VLLM_API_KEY", raising=False) + route: Final = respx_mock.post(f"{_API_BASE}/v1/chat/completions").mock( + return_value=httpx.Response(200, json={"ok": True}) + ) + response: Final = await allm_passthrough_route( + method="POST", + endpoint="v1/chat/completions", + model="hosted_vllm/my-deployment", + api_base=_API_BASE, + json={"model": "anything", "messages": [{"role": "user", "content": "Hello"}]}, + client=AsyncHTTPHandler(), + ) + assert response.status_code == 200 + assert "authorization" not in route.calls[0].request.headers + assert "x-api-key" not in route.calls[0].request.headers + + +@pytest.mark.parametrize( + ("vllm_api_key", "expected_authorization"), + [("sk-vllm-env", "Bearer sk-vllm-env"), ("oidc/env/VLLM_OIDC_TOKEN", "Bearer sk-vllm-oidc")], +) +@pytest.mark.asyncio +async def test_allm_passthrough_route_keyless_hosted_vllm_without_api_base_sends_the_vllm_api_base_key( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch, vllm_api_key: str, expected_authorization: str +) -> None: + monkeypatch.setenv("VLLM_API_BASE", _API_BASE) + monkeypatch.setenv("VLLM_API_KEY", vllm_api_key) + monkeypatch.setenv("VLLM_OIDC_TOKEN", "sk-vllm-oidc") + monkeypatch.delenv("HOSTED_VLLM_API_BASE", raising=False) + monkeypatch.delenv("HOSTED_VLLM_API_KEY", raising=False) + route: Final = respx_mock.post(f"{_API_BASE}/v1/chat/completions").mock( + return_value=httpx.Response(200, json={"ok": True}) + ) + response: Final = await allm_passthrough_route( + method="POST", + endpoint="v1/chat/completions", + model="hosted_vllm/my-deployment", + json={"model": "anything", "messages": [{"role": "user", "content": "Hello"}]}, + client=AsyncHTTPHandler(), + ) + assert response.status_code == 200 + assert route.calls[0].request.headers["authorization"] == expected_authorization + + +@pytest.mark.parametrize( + ("model_api_base", "vllm_api_base"), [(f"{_API_BASE}/", _API_BASE), (_API_BASE, f"{_API_BASE}/")] +) +@pytest.mark.asyncio +async def test_allm_passthrough_route_keyless_hosted_vllm_matches_the_vllm_api_base_across_a_trailing_slash( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch, model_api_base: str, vllm_api_base: str +) -> None: + monkeypatch.setenv("VLLM_API_BASE", vllm_api_base) + monkeypatch.setenv("VLLM_API_KEY", "sk-vllm-env") + monkeypatch.delenv("HOSTED_VLLM_API_BASE", raising=False) + monkeypatch.delenv("HOSTED_VLLM_API_KEY", raising=False) + route: Final = respx_mock.post(url__regex=rf"^{_API_BASE}/+v1/chat/completions$").mock( + return_value=httpx.Response(200, json={"ok": True}) + ) + response: Final = await allm_passthrough_route( + method="POST", + endpoint="v1/chat/completions", + model="hosted_vllm/my-deployment", + api_base=model_api_base, + json={"model": "anything", "messages": [{"role": "user", "content": "Hello"}]}, + client=AsyncHTTPHandler(), + ) + assert response.status_code == 200 + assert route.calls[0].request.headers["authorization"] == "Bearer sk-vllm-env" + + +@pytest.mark.parametrize("vllm_api_key", ["", "oidc/google/audience"]) +@pytest.mark.asyncio +async def test_allm_passthrough_route_unusable_vllm_api_key_sends_no_authorization_header( + respx_mock: respx.MockRouter, monkeypatch: pytest.MonkeyPatch, vllm_api_key: str +) -> None: + monkeypatch.setenv("VLLM_API_BASE", _API_BASE) + monkeypatch.setenv("VLLM_API_KEY", vllm_api_key) + monkeypatch.delenv("HOSTED_VLLM_API_BASE", raising=False) + monkeypatch.delenv("HOSTED_VLLM_API_KEY", raising=False) + route: Final = respx_mock.post(f"{_API_BASE}/v1/chat/completions").mock( + return_value=httpx.Response(200, json={"ok": True}) + ) + response: Final = await allm_passthrough_route( + method="POST", + endpoint="v1/chat/completions", + model="hosted_vllm/my-deployment", + json={"model": "anything", "messages": [{"role": "user", "content": "Hello"}]}, + client=AsyncHTTPHandler(), + ) + assert response.status_code == 200 + assert "authorization" not in route.calls[0].request.headers + + +@pytest.mark.parametrize( + ("api_base", "api_key", "expected_authorization"), + [ + (None, "sk-upstream", "Bearer sk-upstream"), + (None, "sk-hosted-env", "Bearer sk-hosted-env"), + (_API_BASE, "sk-hosted-env", "Bearer sk-hosted-env"), + (None, None, "Bearer sk-hosted-env"), + ], +) +@pytest.mark.asyncio +async def test_allm_passthrough_route_hosted_vllm_on_the_vllm_api_base_sends_its_own_resolved_key( + respx_mock: respx.MockRouter, + monkeypatch: pytest.MonkeyPatch, + api_base: str | None, + api_key: str | None, + expected_authorization: str, +) -> None: + monkeypatch.setenv("VLLM_API_BASE", _API_BASE) + monkeypatch.setenv("VLLM_API_KEY", "sk-vllm-env") + monkeypatch.delenv("HOSTED_VLLM_API_BASE", raising=False) + monkeypatch.setenv("HOSTED_VLLM_API_KEY", "sk-hosted-env") + route: Final = respx_mock.post(f"{_API_BASE}/v1/chat/completions").mock( + return_value=httpx.Response(200, json={"ok": True}) + ) + response: Final = await allm_passthrough_route( + method="POST", + endpoint="v1/chat/completions", + model="hosted_vllm/my-deployment", + api_base=api_base, + api_key=api_key, + json={"model": "anything", "messages": [{"role": "user", "content": "Hello"}]}, + client=AsyncHTTPHandler(), + ) + assert response.status_code == 200 + assert route.calls[0].request.headers["authorization"] == expected_authorization diff --git a/tests/unit/proxy/pass_through_endpoints/test_llm_pass_through_endpoints.py b/tests/unit/proxy/pass_through_endpoints/test_llm_pass_through_endpoints.py index 806d88b2467..aab227e9494 100644 --- a/tests/unit/proxy/pass_through_endpoints/test_llm_pass_through_endpoints.py +++ b/tests/unit/proxy/pass_through_endpoints/test_llm_pass_through_endpoints.py @@ -2759,6 +2759,68 @@ class TestVLLMProxyRoute: mock_factory_route.assert_awaited_once() +@pytest.mark.parametrize( + ("vllm_api_base", "request_path"), + [ + ("http://vllm-upstream.test:8090", "/vllm/v1/chat/completions"), + ("http://vllm-upstream.test:8090/v1", "/vllm/v1/chat/completions"), + ("http://vllm-upstream.test:8090/v1", "/vllm/chat/completions"), + ], +) +def test_vllm_provider_route_reaches_v1_paths_with_its_key_as_a_bearer( + monkeypatch: pytest.MonkeyPatch, vllm_api_base: str, request_path: str +) -> None: + from litellm.proxy import proxy_server + + monkeypatch.setenv("VLLM_API_BASE", vllm_api_base) + monkeypatch.setenv("VLLM_API_KEY", "sk-vllm-upstream") + monkeypatch.setattr(proxy_server, "llm_router", None) + monkeypatch.setattr(litellm, "disable_aiohttp_transport", True) + litellm.in_memory_llm_clients_cache.flush_cache() + monkeypatch.setitem( + proxy_server.app.dependency_overrides, user_api_key_auth, lambda: UserAPIKeyAuth(api_key="sk-virtual") + ) + + with respx.mock(assert_all_called=True) as upstream: + route: Final = upstream.post("http://vllm-upstream.test:8090/v1/chat/completions").mock( + return_value=httpx.Response(200, json={"id": "chatcmpl-1"}) + ) + response: Final = TestClient(proxy_server.app).post( + request_path, + json={"model": "Qwen/Qwen3-0.6B", "messages": [{"role": "user", "content": "hi"}]}, + ) + + assert (response.status_code, response.json()) == (200, {"id": "chatcmpl-1"}) + assert route.calls.last.request.headers["authorization"] == "Bearer sk-vllm-upstream" + assert route.calls.last.request.headers["x-api-key"] == "sk-vllm-upstream" + + +def test_vllm_provider_route_with_an_empty_key_sends_no_credential_headers(monkeypatch: pytest.MonkeyPatch) -> None: + from litellm.proxy import proxy_server + + monkeypatch.setenv("VLLM_API_BASE", "http://vllm-upstream.test:8090") + monkeypatch.setenv("VLLM_API_KEY", "") + monkeypatch.setattr(proxy_server, "llm_router", None) + monkeypatch.setattr(litellm, "disable_aiohttp_transport", True) + litellm.in_memory_llm_clients_cache.flush_cache() + monkeypatch.setitem( + proxy_server.app.dependency_overrides, user_api_key_auth, lambda: UserAPIKeyAuth(api_key="sk-virtual") + ) + + with respx.mock(assert_all_called=True) as upstream: + route: Final = upstream.post("http://vllm-upstream.test:8090/v1/chat/completions").mock( + return_value=httpx.Response(200, json={"id": "chatcmpl-1"}) + ) + response: Final = TestClient(proxy_server.app).post( + "/vllm/v1/chat/completions", + json={"model": "Qwen/Qwen3-0.6B", "messages": [{"role": "user", "content": "hi"}]}, + ) + + assert (response.status_code, response.json()) == (200, {"id": "chatcmpl-1"}) + assert "authorization" not in route.calls.last.request.headers + assert "x-api-key" not in route.calls.last.request.headers + + class TestGigachatProxyRoute: @pytest.mark.asyncio @patch( # test-quality-ok: patching litellm internal for unit test isolation From b195d15b937ab5d09a368ed3972ba3e2cd48d19a Mon Sep 17 00:00:00 2001 From: "devin-ai-integration[bot]" <158243242+devin-ai-integration[bot]@users.noreply.github.com> Date: Fri, 30 Oct 2026 09:28:19 -0700 Subject: [PATCH 09/18] fix(vertex_ai,azure_ai)!: auto-add the compaction, thinking display, and tool change betas on Claude routes (#44747) * fix(vertex_ai): forward and auto-add the compact-2026-09-04 beta on Claude routes * test(vertex_ai): prove the compact-2026-09-04 beta reaches the wire through both Claude handlers * fix(vertex_ai): move a client-sent body compact-2026-09-04 beta to the request header * refactor(vertex_ai): build the Claude beta set without mutation and inject the client in the messages wire test * fix(vertex_ai): keep client body betas in the body and answer 400 on a malformed anthropic_beta * fix(rust): allow the compact-2026-09-04 beta on vertex_ai in the Rust beta table The beta header config now maps compact-2026-09-04 for vertex_ai, so the Rust table that mirrors it must stop rejecting that beta there, and the Rust Vertex messages route adds it for a compaction request like the Python route does. * fix(vertex_ai): keep client body betas out of the Vertex body next to litellm betas, honor drop_params, and send the compaction beta on count_tokens Litellm's own betas replace a client's body anthropic_beta again, as on main, so a body beta Vertex rejects no longer turns a 200 into a 400. A malformed beta drops under drop_params and names its field otherwise. inline-tools-2026-09-15 joins compact-2026-09-04 as header-only, since Vertex rejects both in the body. Vertex count_tokens sends compact-2026-09-04 when the request carries compaction. * fix(logging): log and bill Anthropic messages compaction responses on the Rust route The Rust route hands the served Anthropic body to the logger, which validates it against AnthropicResponse. That model had no compaction content block and no compaction stop reason, so a compaction reply raised and wrote no spend row. * fix(vertex_ai,azure_ai): forward compact-2026-09-04 to Azure AI Foundry, merge Vertex body betas, and redact compaction summaries Azure AI Foundry needs the compact-2026-09-04 header for compaction, a replayed compaction block, and count tokens, so the beta map, the native compaction providers, and the Rust beta table now allow it there. Vertex keeps a client's body betas next to the ones litellm adds, as the Anthropic route does, and turn_off_message_logging now redacts compaction summaries too * fix(vertex_ai,azure_ai): auto-add thinking display and tool change betas, redact streamed compaction summaries Folds the header changes from #44103 into this PR. Vertex AI and Azure AI Claude routes now add thinking-display-updates-2026-08-18 when thinking.display is "updates" and mid-conversation-tool-changes-2026-07-01 when a tool_addition block is present, matching the Anthropic and Bedrock routes. With turn_off_message_logging on, compaction summaries from streamed messages and streamed chat completions are now redacted in spend logs and callback payloads, like the non-streamed response already was. * fix(vertex_ai,azure_ai,rust): add tool change betas on count tokens and Rust messages, bill Rust server tool replies Vertex and Azure count tokens now add mid-conversation-tool-changes-2026-07-01 when the history carries a tool_addition or tool_removal block, through one shared helper with the compaction beta. The Rust /v1/messages route adds thinking-display-updates-2026-08-18 and mid-conversation-tool-changes-2026-07-01 when a request uses them, on the providers that accept each beta. Rust-route logging parses the served body with the Anthropic response parser, so web_search, pause_turn, mcp_tool_use and container_upload replies get a spend row. Legacy enabled thinking keeps display updates on anthropic too, and the Vertex transform reads drop_params through drops_unsupported_params. * style(rust): rewrap beta map test line with cargo fmt * test(rust): expect the display beta on Databricks, which main's beta list now accepts --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> --- .../src/providers/anthropic/beta.rs | 34 +- .../src/anthropic/messages/transformation.rs | 99 +++++- .../src/azure_ai/messages/transformation.rs | 5 + .../src/vertex_ai/messages/transformation.rs | 12 + litellm/anthropic_beta_headers_config.json | 9 +- litellm/anthropic_beta_headers_manager.py | 39 ++- litellm/litellm_core_utils/litellm_logging.py | 15 +- litellm/litellm_core_utils/redact_messages.py | 1 + litellm/llms/anthropic/common_utils.py | 41 ++- .../messages/native/transformation.py | 3 +- .../anthropic/count_tokens/handler.py | 1 + .../anthropic/count_tokens/transformation.py | 8 +- .../anthropic/messages/transformation.py | 30 +- .../anthropic/transformation.py | 82 ++++- .../count_tokens/handler.py | 25 +- litellm/types/llms/anthropic.py | 15 +- .../test_litellm_logging.py | 89 ++++++ .../test_redact_messages.py | 5 +- .../anthropic/test_anthropic_common_utils.py | 61 +++- .../count_tokens/test_token_counter.py | 53 +++ ...azure_anthropic_messages_transformation.py | 88 +++++ .../test_azure_anthropic_transformation.py | 84 +++++ ...artner_models_anthropic_messages_config.py | 195 ++++++++++- ...partner_models_anthropic_transformation.py | 302 +++++++++++++++++- .../count_tokens/test_handler.py | 60 ++++ .../test_anthropic_beta_headers_filtering.py | 51 ++- 26 files changed, 1323 insertions(+), 84 deletions(-) create mode 100644 tests/unit/llms/vertex_ai/vertex_ai_partner_models/count_tokens/test_handler.py diff --git a/litellm-rust/crates/llms-types/src/providers/anthropic/beta.rs b/litellm-rust/crates/llms-types/src/providers/anthropic/beta.rs index 80095cac36c..01ed023999d 100644 --- a/litellm-rust/crates/llms-types/src/providers/anthropic/beta.rs +++ b/litellm-rust/crates/llms-types/src/providers/anthropic/beta.rs @@ -199,7 +199,7 @@ impl AnthropicBeta { &[BetaProvider::BedrockConverse, BetaProvider::Databricks] } Self::ContextManagement20250627 => &[BetaProvider::BedrockConverse], - Self::AdvisorTool20260301 | Self::Compact20260904 => &[ + Self::AdvisorTool20260301 => &[ BetaProvider::AzureAi, BetaProvider::BedrockConverse, BetaProvider::Bedrock, @@ -207,6 +207,12 @@ impl AnthropicBeta { BetaProvider::VertexAi, BetaProvider::Databricks, ], + Self::Compact20260904 => &[ + BetaProvider::BedrockConverse, + BetaProvider::Bedrock, + BetaProvider::BedrockMantle, + BetaProvider::Databricks, + ], Self::Bash20241022 | Self::Bash20250124 | Self::McpServers20251204 @@ -256,14 +262,18 @@ impl AnthropicBeta { Self::InterleavedThinking20250514 | Self::WebSearch20250305 => { &[BetaProvider::BedrockConverse, BetaProvider::Bedrock] } - Self::MidConversationOutputConfig20260701 - | Self::MidConversationToolChanges20260701 => &[ + Self::MidConversationOutputConfig20260701 => &[ BetaProvider::AzureAi, BetaProvider::BedrockConverse, BetaProvider::BedrockMantle, BetaProvider::VertexAi, BetaProvider::Databricks, ], + Self::MidConversationToolChanges20260701 => &[ + BetaProvider::BedrockConverse, + BetaProvider::BedrockMantle, + BetaProvider::Databricks, + ], Self::Output128k20250219 | Self::TokenEfficientTools20250219 => &[ BetaProvider::AzureAi, BetaProvider::BedrockConverse, @@ -483,7 +493,7 @@ mod tests { let config = beta_headers_config(); let config_columns: Vec = config .keys() - .filter(|column| column.as_str() != "description") + .filter(|column| column.as_str() != "description" && !column.ends_with("_header_only")) .cloned() .collect(); let beta_provider_columns: Vec = BetaProvider::VARIANTS @@ -502,12 +512,16 @@ mod tests { let config = beta_headers_config(); for provider in BetaProvider::VARIANTS { for beta in &AnthropicBeta::KNOWN { - let expected = config - .get(provider.as_ref()) - .and_then(serde_json::Value::as_object) - .and_then(|column| column.get(beta.as_str())) - .and_then(serde_json::Value::as_str) - .map(str::to_string); + let cell = |column: &str| { + config + .get(column) + .and_then(serde_json::Value::as_object) + .and_then(|values| values.get(beta.as_str())) + .and_then(serde_json::Value::as_str) + .map(str::to_string) + }; + let expected = + cell(&format!("{provider}_header_only")).or_else(|| cell(provider.as_ref())); let actual = beta.on(*provider).map(|name| name.to_string()); assert_eq!(actual, expected, "provider {provider}, beta {beta}"); if matches!(beta, AnthropicBeta::AdvancedToolUse20251120) diff --git a/litellm-rust/crates/llms/src/anthropic/messages/transformation.rs b/litellm-rust/crates/llms/src/anthropic/messages/transformation.rs index 6657cbf262f..fb6d7259bf9 100644 --- a/litellm-rust/crates/llms/src/anthropic/messages/transformation.rs +++ b/litellm-rust/crates/llms/src/anthropic/messages/transformation.rs @@ -1,7 +1,8 @@ use litellm_auth::CredentialPlacement; use litellm_llms_types::{ formats::messages::{ - ContextEdit, ContextManagement, Message, MessagesOptionalParams, MessagesRequest, Speed, + ContentBlock, ContentBlockType, ContextEdit, ContextManagement, Message, + MessagesOptionalParams, MessagesRequest, Speed, ThinkingConfig, ThinkingDisplay, }, providers::anthropic::{AnthropicBeta, BetaProvider, BetaSet, DEFAULT_HEADERS}, recognized::Recognized, @@ -203,6 +204,10 @@ fn feature_betas(request: &MessagesRequest) -> BetaSet { .then_some(AnthropicBeta::FastMode20260201), messages_carry_output_config(&request.messages) .then_some(AnthropicBeta::PerTurnControl20260701), + uses_thinking_display_updates(params) + .then_some(AnthropicBeta::ThinkingDisplayUpdates20260818), + messages_carry_tool_changes(&request.messages) + .then_some(AnthropicBeta::MidConversationToolChanges20260701), has_advisor_tool(tools).then_some(AnthropicBeta::AdvisorTool20260301), is_tool_search_used(tools).then_some(AnthropicBeta::AdvancedToolUse20251120), ] @@ -253,6 +258,36 @@ fn messages_carry_output_config(messages: &[Message]) -> bool { .any(|message| message.extra.contains_key("output_config")) } +fn uses_thinking_display_updates(params: &MessagesOptionalParams) -> bool { + let display = match params.thinking.as_ref().and_then(Recognized::known) { + Some(ThinkingConfig::Enabled(thinking)) => thinking.display.as_ref(), + Some(ThinkingConfig::Adaptive(thinking)) => thinking.display.as_ref(), + _ => None, + }; + display.and_then(Recognized::known) == Some(&ThinkingDisplay::Updates) +} + +fn is_tool_reference_change(block: &ContentBlock) -> bool { + matches!( + &block.block_type, + Some(ContentBlockType::Other(block_type)) + if block_type == "tool_addition" || block_type == "tool_removal" + ) && block + .extra + .get("tool") + .and_then(|tool| tool.get("type")) + .and_then(Value::as_str) + == Some("tool_reference") +} + +fn messages_carry_tool_changes(messages: &[Message]) -> bool { + messages + .iter() + .filter(|message| message.role == "system") + .flat_map(Message::blocks) + .any(is_tool_reference_change) +} + fn unsupported_param(model: &str, param: &str, value: &str, hint: &str) -> Error { Error::InvalidRequest(crate::ErrorDetail::UnsupportedParameter { model: model.into(), @@ -1071,6 +1106,41 @@ mod tests { json!({"safeguards": [{"type": "dangerous_tool_use", "classifier_context": {"v": 1, "permission_mode": "auto"}}]}), &[] )] + #[case::adaptive_display_updates( + json!({"thinking": {"type": "adaptive", "display": "updates"}}), + &["thinking-display-updates-2026-08-18"] + )] + #[case::enabled_display_updates( + json!({"thinking": {"type": "enabled", "budget_tokens": 2048, "display": "updates"}}), + &["thinking-display-updates-2026-08-18"] + )] + #[case::summarized_display(json!({"thinking": {"type": "adaptive", "display": "summarized"}}), &[])] + #[case::disabled_thinking(json!({"thinking": {"type": "disabled"}}), &[])] + #[case::system_tool_addition( + json!({"messages": [ + {"role": "user", "content": "Use the ping tool"}, + {"role": "system", "content": [{"type": "tool_addition", "tool": {"type": "tool_reference", "name": "ping"}}]}, + ]}), + &["mid-conversation-tool-changes-2026-07-01"] + )] + #[case::system_tool_removal( + json!({"messages": [ + {"role": "system", "content": [{"type": "tool_removal", "tool": {"type": "tool_reference", "name": "ping"}}]}, + ]}), + &["mid-conversation-tool-changes-2026-07-01"] + )] + #[case::user_tool_addition( + json!({"messages": [ + {"role": "user", "content": [{"type": "tool_addition", "tool": {"type": "tool_reference", "name": "ping"}}]}, + ]}), + &[] + )] + #[case::tool_addition_without_a_tool_reference( + json!({"messages": [ + {"role": "system", "content": [{"type": "tool_addition", "tool": {"type": "tool_definition"}}]}, + ]}), + &[] + )] fn feature_betas_follow_the_request(#[case] fields: Value, #[case] expected: &[&str]) { assert_eq!(feature_betas(&request(fields)), betas(expected)); } @@ -1093,6 +1163,33 @@ mod tests { ); } + #[rstest] + #[case::vertex_ai( + BetaProvider::VertexAi, + &["mid-conversation-tool-changes-2026-07-01", "thinking-display-updates-2026-08-18"] + )] + #[case::azure_ai( + BetaProvider::AzureAi, + &["mid-conversation-tool-changes-2026-07-01", "thinking-display-updates-2026-08-18"] + )] + #[case::bedrock_converse(BetaProvider::BedrockConverse, &["thinking-display-updates-2026-08-18"])] + #[case::databricks(BetaProvider::Databricks, &["thinking-display-updates-2026-08-18"])] + fn tool_change_and_display_betas_reach_the_providers_that_accept_them( + #[case] provider: BetaProvider, + #[case] expected: &[&str], + ) { + let fields = json!({ + "thinking": {"type": "adaptive", "display": "updates"}, + "messages": [ + {"role": "system", "content": [{"type": "tool_addition", "tool": {"type": "tool_reference", "name": "ping"}}]}, + ], + }); + assert_eq!( + provider_feature_betas(&request(fields), provider), + betas(expected) + ); + } + #[rstest] fn hosts_without_a_beta_map_column_get_safeguards_without_a_derived_beta() { let client = headers(&[("anthropic-beta", "claude-code-20250219")]); diff --git a/litellm-rust/crates/llms/src/azure_ai/messages/transformation.rs b/litellm-rust/crates/llms/src/azure_ai/messages/transformation.rs index bf934fb4d4c..d578acea68c 100644 --- a/litellm-rust/crates/llms/src/azure_ai/messages/transformation.rs +++ b/litellm-rust/crates/llms/src/azure_ai/messages/transformation.rs @@ -545,6 +545,11 @@ mod tests { &[("anthropic-beta", "web-search-2025-03-05")], &[("x-api-key", "k"), ("anthropic-beta", "structured-outputs-2025-11-13,web-search-2025-03-05")] )] + #[case::compaction_needs_the_compact_2026_09_04_beta( + json!({"compaction": {"type": "summarize"}}), + &[], + &[("x-api-key", "k"), ("anthropic-beta", "compact-2026-09-04")] + )] #[case::no_feature_needs_a_beta(json!({}), &[], &[("x-api-key", "k")])] #[case::safeguards_need_the_dangerous_tool_use_beta( json!({"safeguards": [{"type": "dangerous_tool_use"}]}), diff --git a/litellm-rust/crates/llms/src/vertex_ai/messages/transformation.rs b/litellm-rust/crates/llms/src/vertex_ai/messages/transformation.rs index 46e5a68dcb8..e5e3a056fab 100644 --- a/litellm-rust/crates/llms/src/vertex_ai/messages/transformation.rs +++ b/litellm-rust/crates/llms/src/vertex_ai/messages/transformation.rs @@ -798,6 +798,18 @@ mod tests { ]}}), &["compact-2026-01-12", "context-management-2025-06-27"] )] + #[case::compaction(json!({"compaction": {"type": "summarize"}}), &["compact-2026-09-04"])] + #[case::thinking_display_updates( + json!({"thinking": {"type": "adaptive", "display": "updates"}}), + &["thinking-display-updates-2026-08-18"] + )] + #[case::tool_addition( + json!({"messages": [ + {"role": "user", "content": "Use the ping tool"}, + {"role": "system", "content": [{"type": "tool_addition", "tool": {"type": "tool_reference", "name": "ping"}}]} + ]}), + &["mid-conversation-tool-changes-2026-07-01"] + )] #[case::advisor_tool_is_not_a_vertex_beta( json!({"tools": [{"type": "advisor_20260301", "name": "advisor"}]}), &[] diff --git a/litellm/anthropic_beta_headers_config.json b/litellm/anthropic_beta_headers_config.json index cee5ce15469..253074b0e1f 100644 --- a/litellm/anthropic_beta_headers_config.json +++ b/litellm/anthropic_beta_headers_config.json @@ -1,5 +1,5 @@ { - "description": "Mapping of Anthropic beta headers for each provider. Keys are input header names, values are provider-specific header names (or null if unsupported). Only headers present in mapping keys with non-null values can be forwarded.", + "description": "Mapping of Anthropic beta headers for each provider. Keys are input header names, values are provider-specific header names (or null if unsupported). Only headers present in mapping keys with non-null values can be forwarded. A _header_only column lists betas that provider accepts only as the anthropic-beta HTTP header: they are forwarded there and dropped from the anthropic_beta body field. Proxies that predate these columns ignore them and read the provider column alone.", "anthropic": { "advisor-tool-2026-03-01": "advisor-tool-2026-03-01", "advanced-tool-use-2025-11-20": "advanced-tool-use-2025-11-20", @@ -46,6 +46,7 @@ "bash_20250124": null, "code-execution-2025-08-25": "code-execution-2025-08-25", "compact-2026-01-12": null, + "compact-2026-09-04": "compact-2026-09-04", "computer-use-2025-01-24": "computer-use-2025-01-24", "computer-use-2025-11-24": "computer-use-2025-11-24", "context-1m-2025-08-07": "context-1m-2025-08-07", @@ -59,6 +60,7 @@ "mcp-client-2025-11-20": "mcp-client-2025-11-20", "mcp-client-2025-04-04": "mcp-client-2025-04-04", "mcp-servers-2025-12-04": null, + "mid-conversation-tool-changes-2026-07-01": "mid-conversation-tool-changes-2026-07-01", "output-128k-2025-02-19": null, "structured-output-2024-03-01": null, "per-turn-control-2026-07-01": "per-turn-control-2026-07-01", @@ -202,6 +204,7 @@ "mcp-client-2025-11-20": null, "mcp-client-2025-04-04": null, "mcp-servers-2025-12-04": null, + "mid-conversation-tool-changes-2026-07-01": "mid-conversation-tool-changes-2026-07-01", "output-128k-2025-02-19": null, "structured-output-2024-03-01": null, "per-turn-control-2026-07-01": "per-turn-control-2026-07-01", @@ -218,6 +221,10 @@ "web-search-2025-03-05": "web-search-2025-03-05", "inline-tools-2026-09-15": "inline-tools-2026-09-15" }, + "vertex_ai_header_only": { + "compact-2026-09-04": "compact-2026-09-04", + "inline-tools-2026-09-15": "inline-tools-2026-09-15" + }, "databricks": { "advisor-tool-2026-03-01": null, "advanced-tool-use-2025-11-20": null, diff --git a/litellm/anthropic_beta_headers_manager.py b/litellm/anthropic_beta_headers_manager.py index 07730a0fbf8..be2ff07e088 100644 --- a/litellm/anthropic_beta_headers_manager.py +++ b/litellm/anthropic_beta_headers_manager.py @@ -30,7 +30,7 @@ from collections.abc import Mapping from concurrent.futures import ThreadPoolExecutor from functools import cache from importlib.resources import files -from typing import Final, TypeAlias +from typing import Final, Literal, TypeAlias import httpx from pydantic import TypeAdapter @@ -47,6 +47,9 @@ BetaHeadersConfig: TypeAlias = Mapping[str, str | Mapping[str, str | None]] _BETA_HEADERS_CONFIG_ADAPTER: Final = TypeAdapter(BetaHeadersConfig) _DECODED_JSON: Final = TypeAdapter(object) +BetaPlacement: TypeAlias = Literal["header", "body"] +_HEADER_ONLY_COLUMN_SUFFIX: Final = "_header_only" + class GetAnthropicBetaHeadersConfig: """ @@ -241,6 +244,31 @@ def get_provider_name(provider: str) -> str: return aliases.get(provider, provider) +def _beta_column(name: str) -> Mapping[str, str | None]: + column: Final[Mapping[str, str | None]] = _load_beta_headers_config().get(name, {}) + return column + + +def _header_only_column(canonical_provider: str) -> Mapping[str, str | None]: + return _beta_column(f"{canonical_provider}{_HEADER_ONLY_COLUMN_SUFFIX}") + + +def _provider_beta_mapping(provider: str, placement: BetaPlacement) -> Mapping[str, str | None]: + canonical_provider: Final = get_provider_name(provider) + column: Final = _beta_column(canonical_provider) + header_only: Final = _header_only_column(canonical_provider) + match placement: + case "header": + return {**column, **header_only} + case "body": + return {beta: name for beta, name in column.items() if beta not in header_only} + + +def get_header_only_betas(provider: str) -> frozenset[str]: + header_only: Final = _header_only_column(get_provider_name(provider)) + return frozenset(beta for beta, name in header_only.items() if name is not None) + + def has_beta_headers_config(provider: str) -> bool: config: Final = _bundled_beta_headers_config() aliases: Final = config.get("provider_aliases", {}) @@ -251,6 +279,7 @@ def has_beta_headers_config(provider: str) -> bool: def filter_and_transform_beta_headers( beta_headers: list[str], provider: str, + placement: BetaPlacement = "header", ) -> list[str]: """ Filter and transform beta headers based on provider's mapping configuration. @@ -263,6 +292,7 @@ def filter_and_transform_beta_headers( Args: beta_headers: List of Anthropic beta header values provider: Provider name (e.g., "anthropic", "bedrock", "vertex_ai") + placement: Whether the betas travel as the anthropic-beta header or the anthropic_beta body field Returns: List of filtered and transformed beta headers for the provider @@ -270,11 +300,7 @@ def filter_and_transform_beta_headers( if not beta_headers: return [] - config: Final = _load_beta_headers_config() - provider = get_provider_name(provider) - - # Get the header mapping for this provider - provider_mapping: Final = config.get(provider, {}) + provider_mapping: Final = _provider_beta_mapping(provider, placement) filtered_headers: Final[set[str]] = set() @@ -426,6 +452,7 @@ def update_request_with_filtered_beta( filtered_body_betas: Final = filter_and_transform_beta_headers( beta_headers=existing_body_betas, provider=provider, + placement="body", ) if filtered_body_betas: diff --git a/litellm/litellm_core_utils/litellm_logging.py b/litellm/litellm_core_utils/litellm_logging.py index cfe8c1739c6..26949d9bbfb 100644 --- a/litellm/litellm_core_utils/litellm_logging.py +++ b/litellm/litellm_core_utils/litellm_logging.py @@ -569,6 +569,7 @@ def _resolve_mantle_region_for_cost( _PER_CALL_LITELLM_PARAM_DICTS: Final = frozenset({"metadata", "litellm_metadata"}) _LITELLM_PARAMS: Final = TypeAdapter(dict[str, object]) +_SERVED_ANTHROPIC_BODY: Final = TypeAdapter(dict[str, object]) _REDACTED_RESULT: Final = TypeAdapter(object) _GUARDRAIL_INFORMATION_KEY: Final = "standard_logging_guardrail_information" @@ -5067,13 +5068,15 @@ class Logging(LiteLLMLoggingBaseClass): verbose_logger.warning("LiteLLM: the anthropic_messages stream assembled no response, logging an empty one") return litellm.ModelResponse(model=self.model) else: - from litellm.types.llms.anthropic import AnthropicResponse - - pydantic_result: Final = AnthropicResponse.model_validate({"stop_sequence": None, **(served_body or {})}) - import httpx - + anthropic_body: Final = ( + served_body + if served_body is not None + else _SERVED_ANTHROPIC_BODY.validate_python( + result.model_dump() if isinstance(result, BaseModel) else result + ) + ) result = litellm.AnthropicConfig().transform_parsed_response( - completion_response=pydantic_result.model_dump(), + completion_response={"stop_sequence": None, **anthropic_body}, raw_response=httpx.Response( status_code=200, headers={}, diff --git a/litellm/litellm_core_utils/redact_messages.py b/litellm/litellm_core_utils/redact_messages.py index 548d088d833..981859c620b 100644 --- a/litellm/litellm_core_utils/redact_messages.py +++ b/litellm/litellm_core_utils/redact_messages.py @@ -249,6 +249,7 @@ def _redacted_provider_specific_fields(fields: Mapping[str, object]) -> Mapping[ "reasoningContentBlocks", "citations", "compaction_blocks", + "compaction_start", "compaction_delta", ) if key in fields diff --git a/litellm/llms/anthropic/common_utils.py b/litellm/llms/anthropic/common_utils.py index 6d1ac24af55..e1f9a7d540f 100644 --- a/litellm/llms/anthropic/common_utils.py +++ b/litellm/llms/anthropic/common_utils.py @@ -58,6 +58,7 @@ from litellm.types.llms.anthropic import ( ANTHROPIC_THINKING_BINDING_CONTROLS_BETA_HEADER, ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER, ANTHROPIC_TOOL_SEARCH_TOOL_TYPES, + LEGACY_THINKING_DISPLAY_UPDATES_PROVIDERS, AllAnthropicToolsValues, AnthropicMcpServerTool, AnthropicMessagesToolChoice, @@ -126,6 +127,8 @@ _JSON_VALUE: Final = TypeAdapter(JsonValue) _CLAUDE_CODE_USER_AGENT_PREFIXES: Final = ("claude-cli/", "claude-code/") +NATIVE_COMPACTION_PROVIDERS: Final = frozenset({"anthropic", "azure_ai", "vertex_ai"}) + def is_claude_tool_target(custom_llm_provider: str | None, model: str) -> bool: if custom_llm_provider == "anthropic": @@ -204,7 +207,7 @@ def requires_native_compaction_beta( optional_params: Mapping[str, object], messages: Sequence[object], ) -> bool: - return custom_llm_provider == "anthropic" and ( + return custom_llm_provider in NATIVE_COMPACTION_PROVIDERS and ( optional_params.get("compaction") is not None or carries_signed_compaction_block(messages) ) @@ -246,6 +249,27 @@ def context_management_betas(context_management: JsonValue) -> tuple[str, ...]: ) +def count_tokens_feature_betas( + custom_llm_provider: str, + optional_params: Mapping[str, object], + messages: Sequence[object], +) -> tuple[str, ...]: + return tuple( + beta + for beta, used in ( + ( + ANTHROPIC_BETA_HEADER_VALUES.COMPACT_2026_09_04.value, + requires_native_compaction_beta(custom_llm_provider, optional_params, messages), + ), + ( + ANTHROPIC_MID_CONVERSATION_TOOL_CHANGES_BETA_HEADER, + AnthropicModelInfo().is_mid_conversation_tool_change_used(messages), + ), + ) + if used + ) + + def supports_anthropic_cache_control(model: str, custom_llm_provider: str | None) -> bool: from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider from litellm.utils import supports_prompt_caching @@ -645,10 +669,12 @@ class AnthropicModelInfo(BaseLLMModelInfo): file_ids: Final = get_file_ids_from_messages(messages) return len(file_ids) > 0 - def is_thinking_display_updates_used(self, thinking: AnthropicThinkingParam | None) -> bool: - if not isinstance(thinking, dict): - return False - return thinking.get("type") in ("adaptive", "enabled") and thinking.get("display") == "updates" + def is_thinking_display_updates_used(self, thinking: object) -> bool: + match thinking: + case {"type": "adaptive" | "enabled", "display": "updates"}: + return True + case _: + return False def thinking_betas(self, thinking: AnthropicThinkingParam | None) -> tuple[str, ...]: return ( @@ -1130,7 +1156,10 @@ class AnthropicModelInfo(BaseLLMModelInfo): budget_tokens: Final = _legacy_budget_tokens(thinking.get("budget_tokens", 0), drop_params=drop_params) display: Final = thinking.get("display") - supported_display: Final = {"display": display} if display in ("summarized", "omitted") else {} + keeps_display: Final = display in ("summarized", "omitted") or ( + display == "updates" and custom_llm_provider in LEGACY_THINKING_DISPLAY_UPDATES_PROVIDERS + ) + supported_display: Final = {"display": display} if keeps_display else {} carried_keys: Final = carried_thinking_keys( thinking, rebuilt_keys=frozenset(("type", "budget_tokens", "display")), drop_params=drop_params ) diff --git a/litellm/llms/anthropic/messages/native/transformation.py b/litellm/llms/anthropic/messages/native/transformation.py index c148a0defe8..5281ad99620 100644 --- a/litellm/llms/anthropic/messages/native/transformation.py +++ b/litellm/llms/anthropic/messages/native/transformation.py @@ -41,6 +41,7 @@ from litellm.types.llms.anthropic_tool_search import get_tool_search_beta_header from litellm.types.router import GenericLiteLLMParams from ...common_utils import ( + NATIVE_COMPACTION_PROVIDERS, OBJECT_LIST_ADAPTER, AnthropicError, AnthropicModelInfo, @@ -128,7 +129,7 @@ class AnthropicMessagesConfig(BaseAnthropicMessagesConfig): "tool_choice", "thinking", "context_management", - *(("compaction",) if self._resolved_provider == "anthropic" else ()), + *(("compaction",) if self._resolved_provider in NATIVE_COMPACTION_PROVIDERS else ()), "output_format", "inference_geo", "speed", diff --git a/litellm/llms/azure_ai/anthropic/count_tokens/handler.py b/litellm/llms/azure_ai/anthropic/count_tokens/handler.py index 4b3f263201a..c324f0f0980 100644 --- a/litellm/llms/azure_ai/anthropic/count_tokens/handler.py +++ b/litellm/llms/azure_ai/anthropic/count_tokens/handler.py @@ -84,6 +84,7 @@ class AzureAIAnthropicCountTokensHandler(AzureAIAnthropicCountTokensConfig): headers: Final = self.get_required_headers( api_key=api_key, litellm_params=litellm_params, + messages=messages, ) if not {"x-api-key", "Authorization"} & headers.keys(): raise AnthropicError( diff --git a/litellm/llms/azure_ai/anthropic/count_tokens/transformation.py b/litellm/llms/azure_ai/anthropic/count_tokens/transformation.py index 6e3cfce5eb4..c02c92a95d6 100644 --- a/litellm/llms/azure_ai/anthropic/count_tokens/transformation.py +++ b/litellm/llms/azure_ai/anthropic/count_tokens/transformation.py @@ -4,11 +4,13 @@ Azure AI Anthropic CountTokens API transformation logic. Extends the base Anthropic CountTokens transformation with Azure authentication. """ +from collections.abc import Sequence from typing import Final from urllib.parse import urlsplit from litellm.constants import ANTHROPIC_TOKEN_COUNTING_BETA_VERSION from litellm.litellm_core_utils.url_utils import join_api_base +from litellm.llms.anthropic.common_utils import count_tokens_feature_betas, merge_anthropic_beta_headers from litellm.llms.anthropic.count_tokens.transformation import ( AnthropicCountTokensConfig, ) @@ -28,6 +30,7 @@ class AzureAIAnthropicCountTokensConfig(AnthropicCountTokensConfig): self, api_key: str | None, litellm_params: dict[str, object] | None = None, + messages: Sequence[object] = (), ) -> dict[str, str]: """ Get the required headers for the Azure AI Anthropic CountTokens API. @@ -38,6 +41,7 @@ class AzureAIAnthropicCountTokensConfig(AnthropicCountTokensConfig): Args: api_key: The Azure AI API key litellm_params: Optional LiteLLM parameters for additional auth config + messages: The request messages, read for betas the counted features need Returns: Dictionary of required headers with both x-api-key and Azure authentication @@ -46,7 +50,9 @@ class AzureAIAnthropicCountTokensConfig(AnthropicCountTokensConfig): headers: Final = { "Content-Type": "application/json", "anthropic-version": "2023-06-01", - "anthropic-beta": ANTHROPIC_TOKEN_COUNTING_BETA_VERSION, + "anthropic-beta": merge_anthropic_beta_headers( + ANTHROPIC_TOKEN_COUNTING_BETA_VERSION, count_tokens_feature_betas("azure_ai", {}, messages) + ), **({"x-api-key": api_key} if api_key else {}), } diff --git a/litellm/llms/vertex_ai/vertex_ai_partner_models/anthropic/messages/transformation.py b/litellm/llms/vertex_ai/vertex_ai_partner_models/anthropic/messages/transformation.py index da82766c2b9..98fa8ed2b77 100644 --- a/litellm/llms/vertex_ai/vertex_ai_partner_models/anthropic/messages/transformation.py +++ b/litellm/llms/vertex_ai/vertex_ai_partner_models/anthropic/messages/transformation.py @@ -1,9 +1,15 @@ from collections.abc import Mapping, Sequence from typing import Final +from pydantic import TypeAdapter + from litellm.exceptions import UnsupportedParamsError from litellm.litellm_core_utils.core_helpers import drops_unsupported_params -from litellm.llms.anthropic.common_utils import AnthropicModelInfo, context_management_betas +from litellm.llms.anthropic.common_utils import ( + AnthropicModelInfo, + context_management_betas, + requires_native_compaction_beta, +) from litellm.llms.anthropic.messages.native.transformation import ( AnthropicMessagesConfig, messages_carry_output_config, @@ -11,6 +17,7 @@ from litellm.llms.anthropic.messages.native.transformation import ( from litellm.types.llms.anthropic import ( ANTHROPIC_BETA_HEADER_VALUES, ANTHROPIC_HOSTED_TOOLS, + ANTHROPIC_MID_CONVERSATION_TOOL_CHANGES_BETA_HEADER, ) from litellm.types.llms.anthropic_tool_search import get_tool_search_beta_header from litellm.types.llms.vertex_ai import VertexPartnerProvider @@ -19,6 +26,8 @@ from litellm.types.router import GenericLiteLLMParams from ....vertex_llm_base import VertexBase from ..output_params_utils import sanitize_vertex_anthropic_output_params +_OPTIONAL_PARAMS: Final = TypeAdapter(Mapping[str, object]) + _VERSION_HEADER_VERTEX_READS_FROM_BODY: Final = "anthropic-version" _VERTEX_DEFAULT_VERSION: Final = "vertex-2023-10-16" _VERSIONS_SERVED_AS_VERTEX_DEFAULT: Final = frozenset({"", "2023-01-01", "2023-06-01"}) @@ -190,14 +199,23 @@ class VertexAIPartnerModelsAnthropicMessagesConfig(AnthropicMessagesConfig, Vert if messages_carry_output_config(messages): beta_values.add(ANTHROPIC_BETA_HEADER_VALUES.PER_TURN_CONTROL_2026_07_01.value) - beta_values.update( + typed_params: Final = _OPTIONAL_PARAMS.validate_python(optional_params) + if requires_native_compaction_beta("vertex_ai", typed_params, messages): + beta_values.add(ANTHROPIC_BETA_HEADER_VALUES.COMPACT_2026_09_04.value) + + tool_change_betas: Final = ( + (ANTHROPIC_MID_CONVERSATION_TOOL_CHANGES_BETA_HEADER,) + if anthropic_model_info.is_mid_conversation_tool_change_used(messages) + else () + ) + all_beta_values: Final = beta_values.union( anthropic_model_info.thinking_betas( optional_params.get("thinking"), # pyright: ignore[reportUnknownMemberType, reportUnknownArgumentType] # optional_params is the base config's bare dict - ) + ), + tool_change_betas, ) - - if beta_values: - request_headers["anthropic-beta"] = ",".join(beta_values) + if all_beta_values: + request_headers["anthropic-beta"] = ",".join(sorted(all_beta_values)) return request_headers, api_base diff --git a/litellm/llms/vertex_ai/vertex_ai_partner_models/anthropic/transformation.py b/litellm/llms/vertex_ai/vertex_ai_partner_models/anthropic/transformation.py index 79fd9f9e8b1..3d1690fca33 100644 --- a/litellm/llms/vertex_ai/vertex_ai_partner_models/anthropic/transformation.py +++ b/litellm/llms/vertex_ai/vertex_ai_partner_models/anthropic/transformation.py @@ -2,19 +2,23 @@ ## Handler file for calling claude-3 on vertex ai import base64 import json -from collections.abc import Mapping +from collections.abc import Mapping, Sequence from typing import TYPE_CHECKING, Final import httpx -from pydantic import TypeAdapter +from pydantic import TypeAdapter, ValidationError import litellm +from litellm.anthropic_beta_headers_manager import get_header_only_betas +from litellm.litellm_core_utils.core_helpers import drops_unsupported_params from litellm.litellm_core_utils.prompt_templates.image_handling import RemoteMedia, inline_remote_image_urls from litellm.llms.base_llm.chat.transformation import LiteLLMLoggingObj +from litellm.types.llms.anthropic import ANTHROPIC_BETA_HEADER_VALUES from litellm.types.llms.openai import AllMessageValues from litellm.types.utils import ModelResponse from ....anthropic.chat.transformation import AnthropicConfig +from ....anthropic.common_utils import AnthropicError, requires_native_compaction_beta from .output_params_utils import sanitize_vertex_anthropic_output_params if TYPE_CHECKING: @@ -36,6 +40,32 @@ def _vertex_labels_header(labels: object, header_names: Mapping[str, object]) -> return base64.b64encode(json.dumps(string_labels).encode()).decode() if string_labels else None +_BETA_VALUE: Final[TypeAdapter[str | tuple[str, ...] | None]] = TypeAdapter(str | tuple[str, ...] | None) + +_REQUEST_MESSAGES: Final = TypeAdapter(Sequence[object]) + + +def _beta_names(value: object) -> frozenset[str] | None: + try: + parsed: Final = _BETA_VALUE.validate_python(value) + except ValidationError: + return None + names: Final = parsed.split(",") if isinstance(parsed, str) else parsed or () + return frozenset(name.strip() for name in names if name.strip()) + + +def _client_betas(value: object, field: str, drop_params: bool) -> frozenset[str]: + names: Final = _beta_names(value) + if names is not None: + return names + if drop_params: + return frozenset() + raise AnthropicError( + status_code=400, + message=f"{field} must be a string or a list of strings, got {type(value).__name__}", + ) + + class VertexAIError(Exception): def __init__(self, status_code, message): self.status_code = status_code @@ -157,21 +187,43 @@ class VertexAIAnthropicConfig(AnthropicConfig): if context_management: self._add_context_management_beta_headers(beta_set, context_management) - extra_headers: Final = optional_params.get("extra_headers") or {} - anthropic_beta_value: Final = extra_headers.get("anthropic-beta", "") - if isinstance(anthropic_beta_value, str) and anthropic_beta_value: - for beta in anthropic_beta_value.split(","): - beta = beta.strip() - if beta: - beta_set.add(beta) - elif isinstance(anthropic_beta_value, list): - beta_set.update(anthropic_beta_value) - + request_params: Final = _NON_DEFAULT_PARAMS.validate_python(optional_params) + compaction_betas: Final = ( + frozenset({ANTHROPIC_BETA_HEADER_VALUES.COMPACT_2026_09_04.value}) + if requires_native_compaction_beta( + "vertex_ai", request_params, _REQUEST_MESSAGES.validate_python(data["messages"]) + ) + else frozenset[str]() + ) + drop_params: Final = drops_unsupported_params(litellm_params.get("drop_params")) + extra_headers: Final = _NON_DEFAULT_PARAMS.validate_python(request_params.get("extra_headers") or {}) data.pop("extra_headers", None) + injected_betas: Final = frozenset(beta_set) | _client_betas( + extra_headers.get("anthropic-beta"), "The anthropic-beta header", drop_params + ) + client_body_betas: Final = _client_betas( + _NON_DEFAULT_PARAMS.validate_python(data).get("anthropic_beta"), "anthropic_beta", drop_params + ) - if beta_set: - data["anthropic_beta"] = list(beta_set) - headers["anthropic-beta"] = ",".join(beta_set) + requested_betas: Final = injected_betas | client_body_betas + header_only_betas: Final = get_header_only_betas("vertex_ai") if requested_betas else frozenset[str]() + body_betas: Final = requested_betas - header_only_betas + if body_betas: + data["anthropic_beta"] = sorted(body_betas) + elif "anthropic_beta" in data: + del data["anthropic_beta"] + header_betas: Final = ( + injected_betas + | compaction_betas + | (client_body_betas & header_only_betas) + | _client_betas( + _NON_DEFAULT_PARAMS.validate_python(headers).get("anthropic-beta"), + "The anthropic-beta header", + drop_params, + ) + ) + if header_betas: + headers["anthropic-beta"] = ",".join(sorted(header_betas)) return data diff --git a/litellm/llms/vertex_ai/vertex_ai_partner_models/count_tokens/handler.py b/litellm/llms/vertex_ai/vertex_ai_partner_models/count_tokens/handler.py index 180761f15d9..e341c8f5f48 100644 --- a/litellm/llms/vertex_ai/vertex_ai_partner_models/count_tokens/handler.py +++ b/litellm/llms/vertex_ai/vertex_ai_partner_models/count_tokens/handler.py @@ -6,12 +6,19 @@ Unlike Gemini models which use Google's token counting API, partner models use their respective publisher-specific count-tokens endpoints. """ -from collections.abc import Mapping +from collections.abc import Mapping, Sequence from typing import Any, Final +from pydantic import TypeAdapter + from litellm.litellm_core_utils.core_helpers import drops_unsupported_params from litellm.litellm_core_utils.url_utils import join_api_base -from litellm.llms.anthropic.common_utils import OBJECT_LIST_ADAPTER, is_claude_model_target, tools_without_custom_field +from litellm.llms.anthropic.common_utils import ( + OBJECT_LIST_ADAPTER, + count_tokens_feature_betas, + is_claude_model_target, + tools_without_custom_field, +) from litellm.llms.custom_httpx.http_handler import get_async_httpx_client from litellm.llms.vertex_ai.common_utils import get_vertex_base_url from litellm.llms.vertex_ai.vertex_llm_base import VertexBase @@ -23,6 +30,9 @@ _COUNT_TOKENS_PUBLISHERS: Final = ( ) +_REQUEST_MESSAGES: Final = TypeAdapter(Sequence[object]) + + def count_tokens_request_body( request_data: Mapping[str, object], model: str, drop_params: object ) -> Mapping[str, object]: @@ -43,6 +53,14 @@ def count_tokens_publisher(model: str) -> str | None: return next((publisher for marker, publisher in _COUNT_TOKENS_PUBLISHERS if marker in model), None) +def count_tokens_headers(access_token: str, request_data: Mapping[str, object]) -> Mapping[str, str]: + authorization: Final = {"Authorization": f"Bearer {access_token}"} + betas: Final = count_tokens_feature_betas( + "vertex_ai", request_data, _REQUEST_MESSAGES.validate_python(request_data["messages"]) + ) + return {**authorization, "anthropic-beta": ",".join(betas)} if betas else authorization + + class VertexAIPartnerModelsTokenCounter(VertexBase): """ Token counter for Vertex AI Partner Models. @@ -175,8 +193,7 @@ class VertexAIPartnerModelsTokenCounter(VertexBase): api_base=litellm_params.get("api_base"), ) - # Prepare headers - headers: Final = {"Authorization": f"Bearer {access_token}"} + headers: Final = dict(count_tokens_headers(access_token, request_data)) # Get async HTTP client from litellm import LlmProviders diff --git a/litellm/types/llms/anthropic.py b/litellm/types/llms/anthropic.py index a648376c542..152614e4385 100644 --- a/litellm/types/llms/anthropic.py +++ b/litellm/types/llms/anthropic.py @@ -713,6 +713,14 @@ class AnthropicResponseContentBlockRedactedThinking(LiteLLMBaseModel): data: str +class AnthropicResponseContentBlockCompaction(LiteLLMBaseModel): + model_config = ConfigDict(frozen=True) + + type: Literal["compaction"] + content: str | None = None + signature: str | None = None + + class AnthropicResponseContentBlockServerToolUse(LiteLLMBaseModel): model_config = ConfigDict(frozen=True) @@ -743,7 +751,7 @@ class AnthropicOutputTokensDetails(LiteLLMBaseModel): thinking_tokens: int | None = None -AnthropicFinishReason = Literal["end_turn", "max_tokens", "stop_sequence", "tool_use", "refusal"] +AnthropicFinishReason = Literal["end_turn", "max_tokens", "stop_sequence", "tool_use", "refusal", "compaction"] class AnthropicResponse(LiteLLMBaseModel): @@ -761,6 +769,7 @@ class AnthropicResponse(LiteLLMBaseModel): | AnthropicResponseContentBlockToolUse | AnthropicResponseContentBlockThinking | AnthropicResponseContentBlockRedactedThinking + | AnthropicResponseContentBlockCompaction | AnthropicResponseContentBlockServerToolUse | AnthropicResponseContentBlockWebSearchToolResult ] @@ -808,6 +817,9 @@ class AnthropicThinkingParam(TypedDict, total=False): block_binding: ReadOnly[AnthropicThinkingBlockBinding] +LEGACY_THINKING_DISPLAY_UPDATES_PROVIDERS: Final = frozenset(("anthropic", "azure_ai", "vertex_ai")) + + class ANTHROPIC_HOSTED_TOOLS(str, Enum): WEB_SEARCH = "web_search" BASH = "bash" @@ -828,6 +840,7 @@ class ANTHROPIC_BETA_HEADER_VALUES(str, Enum): CONTEXT_MANAGEMENT_2025_06_27 = "context-management-2025-06-27" COMPACT_2026_01_12 = "compact-2026-01-12" COMPACT_2026_09_04 = "compact-2026-09-04" + INLINE_TOOLS_2026_09_15 = "inline-tools-2026-09-15" STRUCTURED_OUTPUT_2025_09_25 = "structured-outputs-2025-11-13" ADVANCED_TOOL_USE_2025_11_20 = "advanced-tool-use-2025-11-20" FAST_MODE_2026_02_01 = "fast-mode-2026-02-01" diff --git a/tests/unit/litellm_core_utils/test_litellm_logging.py b/tests/unit/litellm_core_utils/test_litellm_logging.py index 336e9f7d24a..d0e31cdd57a 100644 --- a/tests/unit/litellm_core_utils/test_litellm_logging.py +++ b/tests/unit/litellm_core_utils/test_litellm_logging.py @@ -6956,6 +6956,95 @@ def test_handle_anthropic_messages_response_logging_reparses_the_served_body_wit assert logged.usage.completion_tokens == 7 # type: ignore[attr-defined] +@pytest.mark.parametrize( + "compaction_block", + [ + {"type": "compaction", "content": "Summary of the chat so far.", "signature": "sig-abc"}, + {"type": "compaction", "signature": "sig-abc"}, + ], + ids=["with_summary", "without_content"], +) +def test_handle_anthropic_messages_response_logging_reparses_a_served_compaction_response( + compaction_block: dict[str, str], +): + logging_obj = _anthropic_messages_logging_obj() + logging_obj.optional_params = {} + + logged = logging_obj._handle_anthropic_messages_response_logging( + result={ + "id": "msg_compacted", + "type": "message", + "role": "assistant", + "model": "claude-sonnet-5-5", + "content": [compaction_block], + "stop_reason": "compaction", + "stop_sequence": None, + "usage": {"input_tokens": 410, "output_tokens": 95}, + } + ) + + assert isinstance(logged, ModelResponse) + assert logged.id == "msg_compacted" + assert logged.usage.prompt_tokens == 410 + assert logged.usage.completion_tokens == 95 + assert logged.choices[0].message.provider_specific_fields["compaction_blocks"] == [compaction_block] + + +_WEB_SEARCH_REPLY: Final = ( + {"type": "server_tool_use", "id": "srvtoolu_1", "name": "web_search", "input": {"query": "weather"}}, + { + "type": "web_search_tool_result", + "tool_use_id": "srvtoolu_1", + "content": [{"type": "web_search_result", "url": "https://example.com", "title": "Weather"}], + }, + {"type": "text", "text": "It is sunny."}, +) + + +@pytest.mark.parametrize( + ("content", "stop_reason"), + [ + (list(_WEB_SEARCH_REPLY), "end_turn"), + (list(_WEB_SEARCH_REPLY), "pause_turn"), + ([{"type": "text", "text": "The context window is full."}], "model_context_window_exceeded"), + ( + [ + {"type": "mcp_tool_use", "id": "mcptoolu_1", "name": "lookup", "server_name": "docs", "input": {}}, + {"type": "mcp_tool_result", "tool_use_id": "mcptoolu_1", "is_error": False, "content": [{"type": "text", "text": "ok"}]}, + {"type": "text", "text": "Found it."}, + ], + "end_turn", + ), + ([{"type": "container_upload", "file_id": "file_1"}, {"type": "text", "text": "Uploaded."}], "end_turn"), + ], + ids=["web_search", "pause_turn", "model_context_window_exceeded", "mcp_tool_use", "container_upload"], +) +def test_handle_anthropic_messages_response_logging_bills_a_served_server_tool_response( + content: list[dict[str, object]], stop_reason: str +): + logging_obj = _anthropic_messages_logging_obj() + logging_obj.optional_params = {} + + logged = logging_obj._handle_anthropic_messages_response_logging( + result={ + "id": "msg_server_tools", + "type": "message", + "role": "assistant", + "model": "claude-sonnet-5-5", + "content": content, + "stop_reason": stop_reason, + "stop_sequence": None, + "usage": {"input_tokens": 1200, "output_tokens": 80}, + } + ) + + assert isinstance(logged, ModelResponse) + assert logged.id == "msg_server_tools" + assert logged.usage.prompt_tokens == 1200 + assert logged.usage.completion_tokens == 80 + assert logged.choices[0].message.content == content[-1]["text"] + + def test_handle_anthropic_messages_response_logging_passes_model_response_through(): """Anthropic-native path already yields a ModelResponse; it must be returned unchanged.""" logging_obj = _anthropic_messages_logging_obj() diff --git a/tests/unit/litellm_core_utils/test_redact_messages.py b/tests/unit/litellm_core_utils/test_redact_messages.py index 5b64e42d8a8..9e385c588ae 100644 --- a/tests/unit/litellm_core_utils/test_redact_messages.py +++ b/tests/unit/litellm_core_utils/test_redact_messages.py @@ -57,9 +57,10 @@ def _provider_fields() -> dict[str, object]: "thinking_blocks": [{"type": "thinking", "thinking": SECRET}], "reasoningContentBlocks": [{"reasoningText": {"text": SECRET}}], "citations": [{"cited_text": SECRET}], - "thought_signatures": [SIGNATURE], - "compaction_blocks": [{"type": "compaction", "content": SECRET}], + "compaction_blocks": [{"type": "compaction", "content": SECRET, "signature": "compaction-signature"}], + "compaction_start": {"type": "compaction", "content": SECRET}, "compaction_delta": {"type": "compaction_delta", "content": SECRET}, + "thought_signatures": [SIGNATURE], } diff --git a/tests/unit/llms/anthropic/test_anthropic_common_utils.py b/tests/unit/llms/anthropic/test_anthropic_common_utils.py index c1f28e33969..0808ce0ec3c 100644 --- a/tests/unit/llms/anthropic/test_anthropic_common_utils.py +++ b/tests/unit/llms/anthropic/test_anthropic_common_utils.py @@ -3544,12 +3544,11 @@ class TestWifProviderAllowlist: monkeypatch.setenv("WIF_TEST_JWT", "jwt-assertion-value") def test_vertex_anthropic_never_mints_or_sends_the_assertion(self, monkeypatch, wif_engine): + import litellm from litellm.llms.vertex_ai.vertex_ai_partner_models.anthropic.transformation import ( VertexAIAnthropicConfig, ) - import litellm - poster, calls = wif_engine self._env_only_wif(monkeypatch) @@ -3775,7 +3774,7 @@ class TestModelDiscovery: ) monkeypatch.setattr("litellm.module_level_client", client) - with pytest.raises(Exception, match="invalid x-api-key") as exc_info: # noqa: B017, PT011 # the callee raises a bare Exception; match pins the sanitized text + with pytest.raises(Exception, match="invalid x-api-key") as exc_info: # noqa: B017 # the callee raises a bare Exception; match pins the sanitized text AnthropicModelInfo().get_models(api_base="https://api.anthropic.com") assert "invalid x-api-key" in str(exc_info.value) @@ -4238,15 +4237,18 @@ def test_thinking_display_beta_requires_active_thinking(thinking: object, expect @pytest.mark.usefixtures("local_model_cost_map") @pytest.mark.parametrize( - ("display", "expected_thinking"), + ("provider", "display", "expected_thinking"), ( - ("summarized", {"type": "adaptive", "display": "summarized"}), - ("omitted", {"type": "adaptive", "display": "omitted"}), - ("updates", {"type": "adaptive"}), + ("azure_ai", "summarized", {"type": "adaptive", "display": "summarized"}), + ("azure_ai", "omitted", {"type": "adaptive", "display": "omitted"}), + ("azure_ai", "updates", {"type": "adaptive", "display": "updates"}), + ("vertex_ai", "updates", {"type": "adaptive", "display": "updates"}), + ("anthropic", "updates", {"type": "adaptive", "display": "updates"}), + ("databricks", "updates", {"type": "adaptive"}), ), ) def test_shared_legacy_thinking_translation_preserves_supported_display( - display: str, expected_thinking: dict[str, str] + provider: str, display: str, expected_thinking: dict[str, str] ) -> None: from litellm.llms.anthropic.common_utils import AnthropicModelInfo @@ -4257,7 +4259,7 @@ def test_shared_legacy_thinking_translation_preserves_supported_display( AnthropicModelInfo.translate_legacy_thinking_for_adaptive_model( model="claude-opus-5", optional_params=optional_params, - custom_llm_provider="azure_ai", + custom_llm_provider=provider, drop_params=False, ) @@ -4552,6 +4554,47 @@ def test_shared_legacy_thinking_translation_drops_a_non_numeric_budget_under_dro ) +_SIGNED_COMPACTION_BLOCK: Final = {"type": "compaction", "content": "summary so far", "signature": "sig"} + + +@pytest.mark.parametrize( + "provider,expected", + [("anthropic", True), ("vertex_ai", True), ("azure_ai", True), ("bedrock", False), ("bedrock_converse", False)], +) +@pytest.mark.parametrize( + "optional_params,messages", + [ + ({"compaction": {"type": "summarize"}}, [{"role": "user", "content": "hi"}]), + ({}, [{"role": "assistant", "content": [_SIGNED_COMPACTION_BLOCK]}, {"role": "user", "content": "go on"}]), + ], + ids=["compaction_param", "signed_compaction_block"], +) +def test_requires_native_compaction_beta_per_provider( + provider: str, expected: bool, optional_params: dict[str, object], messages: list[object] +) -> None: + from litellm.llms.anthropic.common_utils import requires_native_compaction_beta + + assert requires_native_compaction_beta(provider, optional_params, messages) is expected + + +@pytest.mark.parametrize("provider", ["anthropic", "azure_ai", "vertex_ai"]) +@pytest.mark.parametrize( + "optional_params,messages", + [ + ({"max_tokens": 64}, [{"role": "user", "content": "hi"}]), + ({}, [{"role": "assistant", "content": [{"type": "compaction", "content": "unsigned"}]}]), + ({"compaction": None}, [{"role": "user", "content": "hi"}]), + ], + ids=["no_signal", "unsigned_block", "compaction_none"], +) +def test_requires_native_compaction_beta_is_false_without_a_signal( + provider: str, optional_params: dict[str, object], messages: list[object] +) -> None: + from litellm.llms.anthropic.common_utils import requires_native_compaction_beta + + assert requires_native_compaction_beta(provider, optional_params, messages) is False + + @pytest.mark.parametrize( ("custom_llm_provider", "model", "expected"), [ diff --git a/tests/unit/llms/azure_ai/anthropic/count_tokens/test_token_counter.py b/tests/unit/llms/azure_ai/anthropic/count_tokens/test_token_counter.py index 4913d2b0d1d..7807975be24 100644 --- a/tests/unit/llms/azure_ai/anthropic/count_tokens/test_token_counter.py +++ b/tests/unit/llms/azure_ai/anthropic/count_tokens/test_token_counter.py @@ -215,3 +215,56 @@ async def test_count_uses_the_shared_api_base_chat_would_use(monkeypatch: pytest assert result is not None assert result.total_tokens == 6 assert route.called + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + ("assistant_content", "expected_beta"), + [ + ( + [{"type": "compaction", "content": "Summary of the chat so far.", "signature": "sig-abc"}], + "compact-2026-09-04,token-counting-2024-11-01", + ), + ("Use Recipe and Step tables.", "token-counting-2024-11-01"), + ], + ids=("signed_compaction_block", "plain_turn"), +) +async def test_count_sends_the_compact_beta_only_when_the_history_replays_a_compaction_block( + monkeypatch: pytest.MonkeyPatch, assistant_content: object, expected_beta: str +) -> None: + monkeypatch.setattr(litellm, "disable_aiohttp_transport", True) + with respx.mock: + route: Final = respx.post(COUNT_URL).mock(return_value=httpx.Response(200, json={"input_tokens": 376})) + result: Final = await AzureAIAnthropicTokenCounter().count_tokens( + model_to_use="claude-sonnet-5-5", + messages=[ + {"role": "assistant", "content": assistant_content}, + {"role": "user", "content": "What tables did we pick?"}, + ], + contents=None, + deployment={"litellm_params": {"api_key": "azure-key", "api_base": API_BASE}}, + request_model="azure_ai/claude-sonnet-5-5", + ) + assert result is not None + assert result.total_tokens == 376 + assert route.calls.last.request.headers["anthropic-beta"] == expected_beta + + +@pytest.mark.asyncio +async def test_count_sends_the_tool_change_beta_when_the_history_adds_a_tool(monkeypatch: pytest.MonkeyPatch) -> None: + monkeypatch.setattr(litellm, "disable_aiohttp_transport", True) + with respx.mock: + route: Final = respx.post(COUNT_URL).mock(return_value=httpx.Response(200, json={"input_tokens": 541})) + result: Final = await AzureAIAnthropicTokenCounter().count_tokens( + model_to_use="claude-sonnet-5-5", + messages=[ + {"role": "user", "content": "Use the ping tool once."}, + {"role": "system", "content": [{"type": "tool_addition", "tool": {"type": "tool_reference", "name": "ping"}}]}, + ], + contents=None, + deployment={"litellm_params": {"api_key": "azure-key", "api_base": API_BASE}}, + request_model="azure_ai/claude-sonnet-5-5", + ) + assert result is not None + assert result.total_tokens == 541 + assert route.calls.last.request.headers["anthropic-beta"] == "mid-conversation-tool-changes-2026-07-01,token-counting-2024-11-01" diff --git a/tests/unit/llms/azure_ai/claude/test_azure_anthropic_messages_transformation.py b/tests/unit/llms/azure_ai/claude/test_azure_anthropic_messages_transformation.py index 5e0395f84dc..1c4097ad7e0 100644 --- a/tests/unit/llms/azure_ai/claude/test_azure_anthropic_messages_transformation.py +++ b/tests/unit/llms/azure_ai/claude/test_azure_anthropic_messages_transformation.py @@ -607,3 +607,91 @@ async def test_outgoing_azure_request_keeps_an_object_output_format_under_drop_p body: Final = await _azure_request_body_for_output_format(output_format, drop_params=True) assert body["output_format"] == output_format + + +_SIGNED_COMPACTION_BLOCK: Final = {"type": "compaction", "content": "Summary of the chat so far.", "signature": "sig"} + + +@pytest.mark.asyncio +@pytest.mark.usefixtures("local_beta_headers_config") +@pytest.mark.parametrize( + ("messages", "params"), + [ + ([{"role": "user", "content": "Summarize the chat so far."}], {"compaction": {"type": "summarize"}}), + ( + [{"role": "assistant", "content": [_SIGNED_COMPACTION_BLOCK]}, {"role": "user", "content": "Go on"}], + {}, + ), + ], + ids=["compaction_param", "signed_compaction_block"], +) +async def test_outgoing_azure_compaction_request_carries_the_compact_2026_09_04_beta( + messages: list[object], params: dict[str, object] +) -> None: + """Foundry answers 400 "compaction: this parameter requires anthropic-beta: compact-2026-09-04" for the param + and 400 "Input tag 'compaction' ... does not match" for a replayed block without the header, and 200 with it + (live, claude-sonnet-5-5 on Azure AI Foundry, 2026-10-20).""" + received: Final[list[httpx.Request]] = [] + + def foundry_answers(request: httpx.Request) -> httpx.Response: + received.append(request) + return _foundry_answer(request) + + upstream: Final = AsyncHTTPHandler(transport=httpx.MockTransport(foundry_answers)) + try: + await litellm.anthropic.messages.acreate( + max_tokens=256, + messages=messages, + model="azure_ai/claude-sonnet-5-5", + api_key="test-api-key", + api_base="https://test-resource.services.ai.azure.com", + client=upstream, + **params, + ) + finally: + await upstream.close() + + assert len(received) == 1, received + assert received[0].headers["anthropic-beta"].split(",").count("compact-2026-09-04") == 1, dict(received[0].headers) + + +@pytest.mark.usefixtures("local_model_cost_map") +@pytest.mark.parametrize("feature", ("thinking", "output_config", "tool_addition", "tool_removal", "none")) +@pytest.mark.parametrize("explicit_beta", (False, True)) +def test_messages_feature_beta_headers(feature: str, explicit_beta: bool) -> None: + from typing import Final + + from litellm.llms.azure_ai.anthropic.messages_transformation import AzureAnthropicMessagesConfig + from litellm.types.llms.anthropic import ( + ANTHROPIC_BETA_HEADER_VALUES, + ANTHROPIC_MID_CONVERSATION_TOOL_CHANGES_BETA_HEADER, + ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER, + ) + + beta: Final = ( + ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER + if feature == "thinking" + else ANTHROPIC_BETA_HEADER_VALUES.PER_TURN_CONTROL_2026_07_01.value + if feature == "output_config" + else ANTHROPIC_MID_CONVERSATION_TOOL_CHANGES_BETA_HEADER + ) + message: Final = ( + {"role": "system", "content": [], "output_config": {"effort": "high"}} + if feature == "output_config" + else {"role": "system", "content": [{"type": feature, "tool": {"type": "tool_reference", "name": "ping"}}]} + if feature in ("tool_addition", "tool_removal") + else {"role": "user", "content": "Reply with OK"} + ) + headers, _ = AzureAnthropicMessagesConfig().validate_anthropic_messages_environment( + headers={"anthropic-beta": f"existing-beta,{beta}" if explicit_beta else "existing-beta"}, + model="claude-fable-5-1", + messages=[{"role": "user", "content": "Hello"}, message], + optional_params={"thinking": {"type": "adaptive", "display": "updates"}} if feature == "thinking" else {}, + litellm_params={"vertex_ai_project": "test-project", "vertex_ai_location": "us-east5"}, + api_key="test-key", + api_base="https://example.com/anthropic", + ) + + betas: Final = headers["anthropic-beta"].split(",") + assert betas.count(beta) == int(feature != "none" or explicit_beta) + assert betas.count("existing-beta") == 1 diff --git a/tests/unit/llms/azure_ai/claude/test_azure_anthropic_transformation.py b/tests/unit/llms/azure_ai/claude/test_azure_anthropic_transformation.py index 1179b0dd5da..6ff2a6973a3 100644 --- a/tests/unit/llms/azure_ai/claude/test_azure_anthropic_transformation.py +++ b/tests/unit/llms/azure_ai/claude/test_azure_anthropic_transformation.py @@ -5,10 +5,12 @@ sys.path.insert( 0, os.path.abspath(os.path.join(os.path.dirname(__file__), "../../../../..")) ) +from typing import Final from unittest.mock import MagicMock, patch import pytest +from litellm.anthropic_beta_headers_manager import update_headers_with_filtered_beta from litellm.llms.azure_ai.anthropic.transformation import AzureAnthropicConfig from litellm.types.router import GenericLiteLLMParams @@ -484,3 +486,85 @@ def test_chat_flagged_model_keeps_mid_conversation_system_role_in_place(local_mo "role": "system", "content": [{"type": "text", "text": "Answer with exactly one word."}], } + + +@pytest.mark.parametrize( + ("optional_params", "messages", "expected"), + [ + ({"compaction": {"type": "summarize"}}, [{"role": "user", "content": "Hello"}], True), + ( + {}, + [ + {"role": "assistant", "content": [{"type": "compaction", "content": "Summary.", "signature": "sig"}]}, + {"role": "user", "content": "Go on"}, + ], + True, + ), + ({}, [{"role": "user", "content": "Hello"}], False), + ], + ids=["compaction_param", "signed_compaction_block", "no_compaction"], +) +def test_azure_chat_compaction_sends_the_compact_2026_09_04_beta( + local_beta_headers_config: None, optional_params: dict[str, object], messages: list[object], expected: bool +) -> None: + headers: Final = AzureAnthropicConfig().update_headers_with_optional_anthropic_beta( + headers={}, optional_params=optional_params, messages=messages + ) + + filtered: Final = update_headers_with_filtered_beta(headers=headers, provider="azure_ai") + + assert ("compact-2026-09-04" in filtered.get("anthropic-beta", "").split(",")) is expected + + +@pytest.mark.usefixtures("local_model_cost_map") +@pytest.mark.parametrize("thinking_type", ("adaptive", "enabled")) +@pytest.mark.parametrize("display", ("summarized", "omitted", "updates")) +def test_chat_preserves_thinking_display_and_beta(thinking_type: str, display: str) -> None: + from typing import Final + + from litellm.llms.azure_ai.anthropic.transformation import AzureAnthropicConfig + from litellm.types.llms.anthropic import ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER + + config: Final = AzureAnthropicConfig() + optional_params: Final = config.map_openai_params( + non_default_params={"thinking": {"type": thinking_type, "display": display, "budget_tokens": 2048}}, + optional_params={}, + model="claude-fable-5-1", + drop_params=False, + ) + headers: Final = config.validate_environment( + headers={"anthropic-beta": "existing-beta"}, + model="claude-fable-5-1", + messages=[{"role": "user", "content": "Reply with OK"}], + optional_params=optional_params, + litellm_params={}, + api_key="test-key", + ) + result: Final = config.transform_request( + model="claude-fable-5-1", + messages=[{"role": "user", "content": "Reply with OK"}], + optional_params=optional_params, + litellm_params={}, + headers=headers, + ) + + assert result["thinking"]["type"] == "adaptive" + assert result["thinking"]["display"] == display + assert (ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER in headers["anthropic-beta"].split(",")) == (display == "updates") + assert "existing-beta" in headers["anthropic-beta"].split(",") + + +@pytest.mark.parametrize("thinking", ("enabled", {"type": "bogus"}, {"type": "adaptive", "display": "future-mode"})) +def test_beta_detection_preserves_unrecognized_thinking(thinking: object) -> None: + from typing import Final + + from litellm.types.llms.anthropic import ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER + + params: Final = {"thinking": thinking} + headers: Final = AzureAnthropicConfig().validate_environment( + headers={}, model="claude-fable-5-1", messages=[], optional_params=params, + litellm_params={}, api_key="test-key", + ) + + assert params["thinking"] == thinking + assert ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER not in headers.get("anthropic-beta", "").split(",") diff --git a/tests/unit/llms/vertex_ai/vertex_ai_partner_models/anthropic/test_vertex_ai_partner_models_anthropic_messages_config.py b/tests/unit/llms/vertex_ai/vertex_ai_partner_models/anthropic/test_vertex_ai_partner_models_anthropic_messages_config.py index 7f16957ddc4..fc57dd13302 100644 --- a/tests/unit/llms/vertex_ai/vertex_ai_partner_models/anthropic/test_vertex_ai_partner_models_anthropic_messages_config.py +++ b/tests/unit/llms/vertex_ai/vertex_ai_partner_models/anthropic/test_vertex_ai_partner_models_anthropic_messages_config.py @@ -8,11 +8,13 @@ from threading import get_ident from typing import Final from unittest.mock import AsyncMock, MagicMock, patch +import httpx import pytest import requests import responses import litellm +from litellm.anthropic_beta_headers_manager import update_headers_with_filtered_beta from litellm.llms.custom_httpx.http_handler import AsyncHTTPHandler from litellm.llms.vertex_ai.vertex_ai_partner_models.anthropic.messages.transformation import ( VertexAIPartnerModelsAnthropicMessagesConfig, @@ -115,7 +117,11 @@ def _vertex_client() -> MagicMock: async def _send_to_vertex(client: MagicMock, caller_headers: Mapping[str, str], drop_params: bool) -> None: with responses.RequestsMock(assert_all_requests_are_fired=False) as oauth: - oauth.add(responses.POST, "https://oauth2.googleapis.com/token", json={"access_token": "vertex-token", "expires_in": 3600}) + oauth.add( + responses.POST, + "https://oauth2.googleapis.com/token", + json={"access_token": "vertex-token", "expires_in": 3600}, + ) await litellm.anthropic.messages.acreate( model="vertex_ai/claude-sonnet-5", messages=[{"role": "user", "content": "hi"}], @@ -289,7 +295,12 @@ def test_no_safeguards_leaves_dangerous_tool_use_beta_header_out(): assert "dangerous-tool-use-2026-09-03" not in updated_headers.get("anthropic-beta", "") -def _validate_vertex_headers(client_headers, messages, thinking: Mapping[str, object] | None = None): +def _validate_vertex_headers( + client_headers: dict[str, str], + messages: list[object], + optional_params: dict[str, object] | None = None, + thinking: Mapping[str, object] | None = None, +) -> dict[str, str]: config = VertexAIPartnerModelsAnthropicMessagesConfig() litellm_params = { "vertex_ai_project": "test-project", @@ -305,7 +316,10 @@ def _validate_vertex_headers(client_headers, messages, thinking: Mapping[str, ob headers=client_headers, model="claude-opus-5-5", messages=messages, - optional_params={"max_tokens": 64, **({"thinking": thinking} if thinking is not None else {})}, + optional_params={ + **(optional_params or {"max_tokens": 64}), + **({"thinking": thinking} if thinking is not None else {}), + }, litellm_params=litellm_params, api_base=None, ) @@ -1040,3 +1054,178 @@ def test_vertex_claude_4_8_plus_cost_map_entries_carry_mid_conversation_system_f and info.get("supports_mid_conversation_system") is not True ] assert missing == [] + + +_SIGNED_COMPACTION_BLOCK: Final = {"type": "compaction", "content": "summary so far", "signature": "sig"} + + +@pytest.mark.parametrize( + "client_headers", + [{"anthropic-beta": "compact-2026-09-04"}, {}], + ids=["client_sends_beta", "client_omits_beta"], +) +@pytest.mark.parametrize( + "optional_params,messages", + [ + ({"max_tokens": 64, "compaction": {"type": "summarize"}}, [{"role": "user", "content": "Hello"}]), + ( + {"max_tokens": 64}, + [{"role": "assistant", "content": [_SIGNED_COMPACTION_BLOCK]}, {"role": "user", "content": "Go on"}], + ), + ], + ids=["compaction_param", "signed_compaction_block"], +) +def test_native_compaction_reaches_vertex_with_compact_2026_09_04_beta( + local_beta_headers_config: None, + client_headers: dict[str, str], + optional_params: dict[str, object], + messages: list[object], +) -> None: + """Vertex serves on-demand compaction only behind compact-2026-09-04 (Google Cloud: beta on + https://platform.claude.com/docs/en/build-with-claude/compaction-on-demand, read 2026-10-05) + and answers 400 "compaction: Extra inputs are not permitted" without it, so the beta rides + along with the `compaction` param or a signed compaction block exactly once, whether or not + the client sent it, and survives the Vertex beta filter.""" + headers: Final = _validate_vertex_headers(client_headers, messages, optional_params) + + filtered: Final = update_headers_with_filtered_beta(headers=headers, provider="vertex_ai") + + assert filtered["anthropic-beta"].split(",").count("compact-2026-09-04") == 1 + + +def test_no_compaction_signal_leaves_compact_2026_09_04_beta_out() -> None: + headers: Final = _validate_vertex_headers({}, [{"role": "user", "content": "Hello"}]) + + assert "compact-2026-09-04" not in headers.get("anthropic-beta", "") + + +_VERTEX_MESSAGE_RESPONSE: Final = { + "id": "msg_vrtx_test", + "type": "message", + "role": "assistant", + "model": "claude-sonnet-4-6", + "content": [{"type": "text", "text": "pong"}], + "stop_reason": "end_turn", + "stop_sequence": None, + "usage": {"input_tokens": 5, "output_tokens": 1}, +} + + +@pytest.mark.asyncio +async def test_vertex_messages_request_sends_compact_2026_09_04_beta_on_the_wire( + local_beta_headers_config: None, +) -> None: + """A /v1/messages compaction request to a Vertex Claude deployment reaches Vertex with the beta in the + anthropic-beta request header and never in the body, the shape Vertex accepts (live, 2026-10-05).""" + mock_response: Final = MagicMock() + mock_response.status_code = 200 + mock_response.text = json.dumps(_VERTEX_MESSAGE_RESPONSE) + mock_response.headers = httpx.Headers({}) + mock_response.json.return_value = _VERTEX_MESSAGE_RESPONSE + + client: Final = MagicMock(spec=AsyncHTTPHandler) + client.post = AsyncMock(return_value=mock_response) + + with patch( + "litellm.llms.vertex_ai.vertex_llm_base.VertexBase._ensure_access_token_async", + AsyncMock(return_value=("token", "test-project")), + ): + await litellm.anthropic.messages.acreate( + model="vertex_ai/claude-sonnet-4-6", + max_tokens=64, + messages=[{"role": "user", "content": "Reply with the single word: pong"}], + compaction={"type": "summarize"}, + vertex_project="test-project", + vertex_location="us-east5", + client=client, + ) + + sent: Final = client.post.call_args.kwargs + sent_body: Final = sent["json"] if "json" in sent else json.loads(sent["data"]) + assert sent["headers"]["anthropic-beta"].split(",").count("compact-2026-09-04") == 1 + assert sent_body["compaction"] == {"type": "summarize"} + assert "anthropic_beta" not in sent_body + + +@pytest.mark.usefixtures("local_model_cost_map", "local_beta_headers_config") +@pytest.mark.parametrize("feature", ("thinking", "output_config", "tool_addition", "tool_removal", "none")) +@pytest.mark.parametrize("explicit_beta", (False, True)) +def test_messages_feature_beta_headers(feature: str, explicit_beta: bool) -> None: + from typing import Final + from unittest.mock import patch + + from google.oauth2.credentials import Credentials + + from litellm.anthropic_beta_headers_manager import update_headers_with_filtered_beta + from litellm.llms.vertex_ai.vertex_ai_partner_models.anthropic.messages.transformation import ( + VertexAIPartnerModelsAnthropicMessagesConfig, + ) + from litellm.types.llms.anthropic import ( + ANTHROPIC_BETA_HEADER_VALUES, + ANTHROPIC_MID_CONVERSATION_TOOL_CHANGES_BETA_HEADER, + ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER, + ) + + beta: Final = ( + ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER + if feature == "thinking" + else ANTHROPIC_BETA_HEADER_VALUES.PER_TURN_CONTROL_2026_07_01.value + if feature == "output_config" + else ANTHROPIC_MID_CONVERSATION_TOOL_CHANGES_BETA_HEADER + ) + message: Final = ( + {"role": "system", "content": [], "output_config": {"effort": "high"}} + if feature == "output_config" + else {"role": "system", "content": [{"type": feature, "tool": {"type": "tool_reference", "name": "ping"}}]} + if feature in ("tool_addition", "tool_removal") + else {"role": "user", "content": "Reply with OK"} + ) + with patch( + "google.auth.default", + return_value=(MagicMock(spec=Credentials, token="test-token", expired=False), "test-project"), + ): + headers, _ = VertexAIPartnerModelsAnthropicMessagesConfig().validate_anthropic_messages_environment( + headers={"anthropic-beta": f"existing-beta,{beta}" if explicit_beta else "existing-beta"}, + model="claude-fable-5-1", + messages=[{"role": "user", "content": "Hello"}, message], + optional_params={"thinking": {"type": "adaptive", "display": "updates"}} if feature == "thinking" else {}, + litellm_params={"vertex_ai_project": "test-project", "vertex_ai_location": "us-east5"}, + api_key="test-key", + api_base="https://example.com/anthropic", + ) + + betas: Final = headers["anthropic-beta"].split(",") + assert betas.count(beta) == int(feature != "none" or explicit_beta) + assert betas.count("existing-beta") == 1 + + if feature in ("tool_addition", "tool_removal"): + filtered_headers: Final = update_headers_with_filtered_beta(headers=headers, provider="vertex_ai") + assert filtered_headers.get("anthropic-beta", "").split(",").count(beta) == 1 + + +@pytest.mark.parametrize("thinking", ("enabled", {"type": "bogus"}, {"type": "adaptive", "display": "future-mode"})) +def test_beta_detection_preserves_unrecognized_thinking(thinking: object) -> None: + from typing import Final + from unittest.mock import patch + + from google.oauth2.credentials import Credentials + + from litellm.types.llms.anthropic import ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER + + params: Final = {"thinking": thinking} + with patch( + "google.auth.default", + return_value=(MagicMock(spec=Credentials, token="test-token", expired=False), "test-project"), + ): + headers, _ = VertexAIPartnerModelsAnthropicMessagesConfig().validate_anthropic_messages_environment( + headers={}, + model="claude-fable-5-1", + messages=[], + optional_params=params, + litellm_params={"vertex_ai_project": "test-project", "vertex_ai_location": "us-east5"}, + api_key="test-key", + api_base="https://example.com/anthropic", + ) + + assert params["thinking"] == thinking + assert ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER not in headers.get("anthropic-beta", "").split(",") diff --git a/tests/unit/llms/vertex_ai/vertex_ai_partner_models/anthropic/test_vertex_ai_partner_models_anthropic_transformation.py b/tests/unit/llms/vertex_ai/vertex_ai_partner_models/anthropic/test_vertex_ai_partner_models_anthropic_transformation.py index 3ae208bb3a8..0a582c95749 100644 --- a/tests/unit/llms/vertex_ai/vertex_ai_partner_models/anthropic/test_vertex_ai_partner_models_anthropic_transformation.py +++ b/tests/unit/llms/vertex_ai/vertex_ai_partner_models/anthropic/test_vertex_ai_partner_models_anthropic_transformation.py @@ -4,7 +4,7 @@ import copy import json from collections.abc import Mapping from typing import Final -from unittest.mock import MagicMock +from unittest.mock import MagicMock, patch import httpx import pytest @@ -12,10 +12,12 @@ import respx from google.oauth2.credentials import Credentials import litellm +from litellm.llms.custom_httpx.http_handler import HTTPHandler from litellm.anthropic_beta_headers_manager import ( update_headers_with_filtered_beta, update_request_with_filtered_beta, ) +from litellm.llms.anthropic.common_utils import AnthropicError from litellm.llms.vertex_ai.vertex_ai_partner_models.anthropic.transformation import ( VertexAIAnthropicConfig, ) @@ -475,8 +477,10 @@ def test_vertex_ai_anthropic_extra_headers_beta_propagation(): def test_vertex_ai_anthropic_inline_tools_beta_survives_the_chat_beta_filter(local_beta_headers_config: None) -> None: """The chat path runs the Vertex beta filter over both the header and the `anthropic_beta` body field right - before the request goes out, so inline-tools-2026-09-15 must survive both (source and date in - tests/unit/test_anthropic_beta_headers_filtering.py).""" + before the request goes out, so inline-tools-2026-09-15 must survive it as the header (source and date in + tests/unit/test_anthropic_beta_headers_filtering.py) and stay out of the body field, which Vertex rejects + with 400 "Unexpected value(s) `inline-tools-2026-09-15` for the `anthropic_beta` parameter" (live, global, + claude-sonnet-5-5, 2026-10-20).""" config: Final = VertexAIAnthropicConfig() headers: Final[dict[str, str]] = {} optional_params: Final[dict[str, object]] = { @@ -497,7 +501,7 @@ def test_vertex_ai_anthropic_inline_tools_beta_survives_the_chat_beta_filter(loc ) assert filtered_headers["anthropic-beta"].split(",").count("inline-tools-2026-09-15") == 1 - assert "inline-tools-2026-09-15" in filtered_request["anthropic_beta"] + assert "inline-tools-2026-09-15" not in filtered_request.get("anthropic_beta", []) def test_vertex_ai_anthropic_extra_headers_beta_merged_with_auto_betas(): @@ -1232,3 +1236,293 @@ def test_refusal_stop_details_kept_through_vertex_anthropic_transform(): provider_fields: Final = result.choices[0].message.provider_specific_fields assert provider_fields["stop_reason"] == "refusal" assert provider_fields["stop_details"] == ANTHROPIC_DOCS_REFUSAL_MESSAGE["stop_details"] + + +_SIGNED_COMPACTION_BLOCK: Final = {"type": "compaction", "content": "summary so far", "signature": "sig"} +_CLEAR_TOOL_USES: Final = {"edits": [{"type": "clear_tool_uses_20250919"}]} + + +def _vertex_request(optional_params: dict[str, object], headers: dict[str, str], litellm_params: dict[str, object] | None = None) -> dict: + return VertexAIAnthropicConfig().transform_request( + model="claude-opus-4-6", + messages=[{"role": "user", "content": "Hello"}], + optional_params={"max_tokens": 100, "is_vertex_request": True, **optional_params}, + litellm_params=litellm_params or {}, + headers=headers, + ) + + +@pytest.mark.parametrize( + "optional_params,messages", + [ + ( + {"max_tokens": 100, "is_vertex_request": True, "compaction": {"type": "summarize"}}, + [{"role": "user", "content": "Hello"}], + ), + ( + {"max_tokens": 100, "is_vertex_request": True}, + [ + { + "role": "assistant", + "content": "", + "provider_specific_fields": {"compaction_blocks": [_SIGNED_COMPACTION_BLOCK]}, + }, + {"role": "user", "content": "Go on"}, + ], + ), + ], + ids=["compaction_param", "signed_compaction_block_sent_back"], +) +def test_vertex_ai_anthropic_native_compaction_adds_compact_2026_09_04_beta( + local_beta_headers_config: None, optional_params: dict[str, object], messages: list[dict[str, object]] +) -> None: + """The chat path skips the Anthropic beta injection for Vertex requests, so the Vertex config + adds compact-2026-09-04 itself (Google Cloud: beta on + https://platform.claude.com/docs/en/build-with-claude/compaction-on-demand, read 2026-10-05) + as the HTTP header only, since Vertex answered 400 "Unexpected value(s) `compact-2026-09-04` + for the `anthropic_beta` parameter" with the name in the body field (us-east5, + claude-sonnet-4-6, 2026-10-05), and the header must survive the Vertex beta filter.""" + headers: Final[dict[str, str]] = {} + + result: Final = VertexAIAnthropicConfig().transform_request( + model="claude-opus-4-6", + messages=messages, + optional_params=optional_params, + litellm_params={}, + headers=headers, + ) + filtered_headers, filtered_body = update_request_with_filtered_beta( + headers=headers, request_data=result, provider="vertex_ai" + ) + + assert filtered_headers["anthropic-beta"].split(",").count("compact-2026-09-04") == 1 + assert "compact-2026-09-04" not in filtered_body.get("anthropic_beta", []) + + +def test_vertex_ai_anthropic_no_compaction_signal_leaves_compact_2026_09_04_beta_out() -> None: + headers: Final[dict[str, str]] = {} + + result: Final = _vertex_request({}, headers) + + assert "compact-2026-09-04" not in result.get("anthropic_beta", []) + assert "compact-2026-09-04" not in headers.get("anthropic-beta", "") + + +def test_vertex_ai_anthropic_client_sent_compact_2026_09_04_stays_out_of_the_body_field( + local_beta_headers_config: None, +) -> None: + headers: Final[dict[str, str]] = {} + + result: Final = _vertex_request( + { + "compaction": {"type": "summarize"}, + "extra_headers": {"anthropic-beta": "compact-2026-09-04,interleaved-thinking-2025-05-14"}, + }, + headers, + ) + + assert headers["anthropic-beta"].split(",").count("compact-2026-09-04") == 1 + assert result["anthropic_beta"] == ["interleaved-thinking-2025-05-14"] + + +def test_vertex_ai_anthropic_compaction_keeps_the_betas_the_client_sent_as_headers( + local_beta_headers_config: None, +) -> None: + headers: Final = {"anthropic-beta": "context-1m-2025-08-07"} + + result: Final = _vertex_request({"compaction": {"type": "summarize"}}, headers) + + assert sorted(headers["anthropic-beta"].split(",")) == ["compact-2026-09-04", "context-1m-2025-08-07"] + assert "anthropic_beta" not in result + + +@pytest.mark.parametrize("header_only_beta", ["compact-2026-09-04", "inline-tools-2026-09-15"]) +def test_vertex_ai_anthropic_client_sent_body_header_only_beta_moves_to_the_header( + local_beta_headers_config: None, header_only_beta: str +) -> None: + """Vertex rejects these betas inside the body `anthropic_beta` field with 400 "Unexpected value(s)" and + accepts them as the header (live, global, claude-sonnet-5-5, 2026-10-20), so a client that opts in through that field still + gets the beta as a header and keeps its other betas in the body.""" + headers: Final[dict[str, str]] = {} + + result: Final = _vertex_request( + {"anthropic_beta": [header_only_beta, "interleaved-thinking-2025-05-14"]}, headers + ) + + assert headers["anthropic-beta"] == header_only_beta + assert result["anthropic_beta"] == ["interleaved-thinking-2025-05-14"] + + +@pytest.mark.parametrize( + "client_beta,body,header", + [ + ( + "context-1m-2025-08-07", + ["context-1m-2025-08-07", "context-management-2025-06-27"], + "context-management-2025-06-27", + ), + ( + "inline-tools-2026-09-15", + ["context-management-2025-06-27"], + "context-management-2025-06-27,inline-tools-2026-09-15", + ), + ], + ids=["body_beta", "header_only_beta"], +) +def test_vertex_ai_anthropic_litellm_betas_merge_with_the_client_body_betas( + local_beta_headers_config: None, client_beta: str, body: list[str], header: str +) -> None: + headers: Final[dict[str, str]] = {} + + result: Final = _vertex_request({"context_management": _CLEAR_TOOL_USES, "anthropic_beta": [client_beta]}, headers) + + assert result["anthropic_beta"] == body + assert headers["anthropic-beta"] == header + + +@pytest.mark.parametrize( + "anthropic_beta", + [["interleaved-thinking-2025-05-14"], ("interleaved-thinking-2025-05-14",)], + ids=["list", "sdk_tuple"], +) +def test_vertex_ai_anthropic_client_sent_body_betas_stay_out_of_the_header(anthropic_beta: object) -> None: + headers: Final[dict[str, str]] = {} + + result: Final = _vertex_request({"anthropic_beta": anthropic_beta}, headers) + + assert result["anthropic_beta"] == ["interleaved-thinking-2025-05-14"] + assert "anthropic-beta" not in headers + + +_MALFORMED_BETAS: Final = pytest.mark.parametrize( + "malformed", [5, [1, 2], {"beta": "x"}], ids=["int", "list_of_ints", "dict"] +) +_MALFORMED_BETA_SOURCES: Final = pytest.mark.parametrize( + "source,field", + [("anthropic_beta", "anthropic_beta"), ("extra_headers", "The anthropic-beta header")], + ids=["body_field", "extra_header"], +) + + +def _malformed_beta_params(source: str, malformed: object) -> dict[str, object]: + if source == "extra_headers": + return {"extra_headers": {"anthropic-beta": malformed}} + return {"anthropic_beta": malformed} + + +@_MALFORMED_BETAS +@_MALFORMED_BETA_SOURCES +def test_vertex_ai_anthropic_malformed_anthropic_beta_answers_400_naming_its_field( + malformed: object, source: str, field: str +) -> None: + with pytest.raises(AnthropicError) as excinfo: + _vertex_request(_malformed_beta_params(source, malformed), {}) + + assert excinfo.value.status_code == 400 + assert excinfo.value.message.startswith(f"{field} must be") + + +@_MALFORMED_BETAS +@_MALFORMED_BETA_SOURCES +@pytest.mark.parametrize("auto_beta", [False, True], ids=["alone", "next_to_litellm_beta"]) +def test_vertex_ai_anthropic_malformed_anthropic_beta_is_dropped_under_drop_params( + local_beta_headers_config: None, malformed: object, source: str, field: str, auto_beta: bool +) -> None: + headers: Final[dict[str, str]] = {} + params: Final = {**_malformed_beta_params(source, malformed), **({"context_management": _CLEAR_TOOL_USES} if auto_beta else {})} + + result: Final = _vertex_request(params, headers, litellm_params={"drop_params": True}) + + assert result.get("anthropic_beta") == (["context-management-2025-06-27"] if auto_beta else None) + assert headers.get("anthropic-beta") == ("context-management-2025-06-27" if auto_beta else None) + + +_VERTEX_MESSAGE_RESPONSE: Final = { + "id": "msg_vrtx_test", + "type": "message", + "role": "assistant", + "model": "claude-sonnet-4-6", + "content": [{"type": "text", "text": "pong"}], + "stop_reason": "end_turn", + "stop_sequence": None, + "usage": {"input_tokens": 5, "output_tokens": 1}, +} + + +def test_vertex_ai_anthropic_completion_sends_compact_2026_09_04_beta_on_the_wire(local_beta_headers_config: None) -> None: + """A chat completion with `compaction` on a Vertex Claude deployment reaches Vertex with the beta in the + anthropic-beta request header and never in the body field, the shape Vertex accepts (live, 2026-10-05).""" + mock_response: Final = MagicMock() + mock_response.status_code = 200 + mock_response.text = json.dumps(_VERTEX_MESSAGE_RESPONSE) + mock_response.headers = {} + mock_response.json.return_value = _VERTEX_MESSAGE_RESPONSE + client: Final = MagicMock(spec=HTTPHandler) + client.post.return_value = mock_response + + with patch( + "litellm.llms.vertex_ai.vertex_llm_base.VertexBase._ensure_access_token", + return_value=("token", "test-project"), + ): + litellm.completion( + model="vertex_ai/claude-sonnet-4-6", + messages=[{"role": "user", "content": "Reply with the single word: pong"}], + compaction={"type": "summarize"}, + vertex_project="test-project", + vertex_location="us-east5", + client=client, + ) + + sent: Final = client.post.call_args.kwargs + sent_body: Final = sent["json"] if "json" in sent else json.loads(sent["data"]) + assert sent["headers"]["anthropic-beta"].split(",").count("compact-2026-09-04") == 1 + assert sent_body["compaction"] == {"type": "summarize"} + assert "anthropic_beta" not in sent_body + + +@pytest.mark.usefixtures("local_model_cost_map") +@pytest.mark.parametrize("thinking_type", ("adaptive", "enabled")) +@pytest.mark.parametrize("display", ("summarized", "omitted", "updates")) +def test_chat_preserves_thinking_display_and_beta(thinking_type: str, display: str) -> None: + from typing import Final + + from litellm.llms.vertex_ai.vertex_ai_partner_models.anthropic.transformation import VertexAIAnthropicConfig + from litellm.types.llms.anthropic import ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER + + config: Final = VertexAIAnthropicConfig() + optional_params: Final = config.map_openai_params( + non_default_params={"thinking": {"type": thinking_type, "display": display, "budget_tokens": 2048}}, + optional_params={}, + model="claude-fable-5-1", + drop_params=False, + ) + optional_params["extra_headers"] = {"anthropic-beta": "existing-beta"} + headers: Final = {} + result: Final = config.transform_request( + model="claude-fable-5-1", + messages=[{"role": "user", "content": "Reply with OK"}], + optional_params=optional_params, + litellm_params={}, + headers=headers, + ) + + assert result["thinking"]["type"] == "adaptive" + assert result["thinking"]["display"] == display + assert (ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER in headers["anthropic-beta"].split(",")) == (display == "updates") + assert "existing-beta" in headers["anthropic-beta"].split(",") + + +@pytest.mark.parametrize("thinking", ("enabled", {"type": "bogus"}, {"type": "adaptive", "display": "future-mode"})) +def test_beta_detection_preserves_unrecognized_thinking(thinking: object) -> None: + from typing import Final + + from litellm.types.llms.anthropic import ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER + + headers: Final = {} + result: Final = VertexAIAnthropicConfig().transform_request( + model="claude-fable-5-1", messages=[{"role": "user", "content": "Hello"}], + optional_params={"thinking": thinking}, litellm_params={}, headers=headers, + ) + + assert result["thinking"] == thinking + assert ANTHROPIC_THINKING_DISPLAY_UPDATES_BETA_HEADER not in headers.get("anthropic-beta", "").split(",") diff --git a/tests/unit/llms/vertex_ai/vertex_ai_partner_models/count_tokens/test_handler.py b/tests/unit/llms/vertex_ai/vertex_ai_partner_models/count_tokens/test_handler.py new file mode 100644 index 00000000000..c03f08ab610 --- /dev/null +++ b/tests/unit/llms/vertex_ai/vertex_ai_partner_models/count_tokens/test_handler.py @@ -0,0 +1,60 @@ +from typing import Final + +import pytest + +from litellm.llms.vertex_ai.vertex_ai_partner_models.count_tokens.handler import count_tokens_headers + +_HISTORY: Final = ({"role": "user", "content": "Hello"},) + + +@pytest.mark.parametrize( + "request_data", + [ + {"messages": [{"role": "assistant", "content": [{"type": "compaction", "content": "so far", "signature": "sig"}]}]}, + {"messages": list(_HISTORY), "compaction": {"type": "summarize"}}, + ], + ids=["signed_compaction_block", "compaction_param"], +) +def test_count_tokens_on_a_compaction_request_sends_the_compact_beta(request_data: dict[str, object]) -> None: + assert count_tokens_headers("token", request_data) == { + "Authorization": "Bearer token", + "anthropic-beta": "compact-2026-09-04", + } + + +@pytest.mark.parametrize( + "request_data", + [ + {"messages": list(_HISTORY)}, + {"messages": [{"role": "assistant", "content": [{"type": "compaction", "content": "so far", "signature": ""}]}]}, + ], + ids=["plain", "unsigned_compaction_block"], +) +def test_count_tokens_without_compaction_sends_no_beta(request_data: dict[str, object]) -> None: + assert count_tokens_headers("token", request_data) == {"Authorization": "Bearer token"} + + +_TOOL_ADDITION: Final = { + "role": "system", + "content": [{"type": "tool_addition", "tool": {"type": "tool_reference", "name": "ping"}}], +} + + +@pytest.mark.parametrize( + ("request_data", "expected_beta"), + [ + ({"messages": [*_HISTORY, _TOOL_ADDITION]}, "mid-conversation-tool-changes-2026-07-01"), + ( + {"messages": [*_HISTORY, _TOOL_ADDITION], "compaction": {"type": "summarize"}}, + "compact-2026-09-04,mid-conversation-tool-changes-2026-07-01", + ), + ], + ids=["tool_addition", "tool_addition_with_compaction"], +) +def test_count_tokens_on_a_tool_change_request_sends_the_tool_change_beta( + request_data: dict[str, object], expected_beta: str +) -> None: + assert count_tokens_headers("token", request_data) == { + "Authorization": "Bearer token", + "anthropic-beta": expected_beta, + } diff --git a/tests/unit/test_anthropic_beta_headers_filtering.py b/tests/unit/test_anthropic_beta_headers_filtering.py index deba7fe9587..c8fef95cf98 100644 --- a/tests/unit/test_anthropic_beta_headers_filtering.py +++ b/tests/unit/test_anthropic_beta_headers_filtering.py @@ -85,7 +85,8 @@ class TestAnthropicBetaHeadersFiltering: filtered = filter_and_transform_beta_headers( beta_headers=all_headers, provider=provider ) - assert ("compact-2026-09-04" in filtered) is (provider == "anthropic") + assert ("compact-2026-09-04" in filtered) is (provider in ("anthropic", "azure_ai", "vertex_ai")) + assert ("mid-conversation-tool-changes-2026-07-01" in filtered) is (provider != "bedrock_converse") for header in unsupported_headers: assert ( @@ -406,15 +407,49 @@ class TestAnthropicBetaHeadersFiltering: test_case["expected"] in filtered ), f"Header '{test_case['input']}' should be mapped to '{test_case['expected']}' for {test_case['provider']}, but got: {filtered}" - def test_filter_and_transform_beta_headers_vertex_ai_keeps_compact(self): - """Vertex AI supports compact context edits, so the compact beta header - must be forwarded instead of stripped (it was previously mapped to null, - which broke compact_20260112 context edits over /v1/messages).""" - filtered = filter_and_transform_beta_headers( - beta_headers=["compact-2026-01-12"], provider="vertex_ai" + @pytest.mark.parametrize("beta", ["compact-2026-09-04", "inline-tools-2026-09-15"]) + def test_vertex_ai_header_only_beta_is_forwarded_as_the_header_and_dropped_from_the_body( + self, beta: str + ) -> None: + headers, body = update_request_with_filtered_beta( + headers={"anthropic-beta": f"{beta},context-1m-2025-08-07"}, + request_data={"anthropic_beta": [beta, "context-1m-2025-08-07"]}, + provider="vertex_ai", ) - assert filtered == ["compact-2026-01-12"] + assert set(headers["anthropic-beta"].split(",")) == {beta, "context-1m-2025-08-07"} + assert body["anthropic_beta"] == ["context-1m-2025-08-07"] + + def test_header_only_betas_stay_out_of_the_provider_columns_older_proxies_read(self) -> None: + """Proxies released before header-only columns existed fetch this map from main, read only the + provider column, and copy its betas into the body field too, where Vertex answers 400. So a + header-only beta may sit in its provider column only if older proxies already forwarded it.""" + header_only_columns: Final = { + column.removesuffix("_header_only"): betas + for column, betas in self.config.items() + if column.endswith("_header_only") + } + forwarded_by_older_proxies: Final = { + provider: {beta for beta in betas if self.config[provider].get(beta) is not None} + for provider, betas in header_only_columns.items() + } + + assert "vertex_ai" in header_only_columns + assert {provider: betas for provider, betas in forwarded_by_older_proxies.items() if betas} == { + "vertex_ai": {"inline-tools-2026-09-15"} + } + + def test_filter_and_transform_beta_headers_vertex_ai_keeps_compact(self): + """Vertex AI serves both compaction betas (threshold compaction behind compact-2026-01-12 + and on-demand compaction behind compact-2026-09-04, Google Cloud: beta on + https://platform.claude.com/docs/en/build-with-claude/compaction-on-demand, read 2026-10-05), + so the Vertex filter forwards both instead of stripping them: a dropped header makes + Vertex answer 400 "compaction: Extra inputs are not permitted".""" + filtered = filter_and_transform_beta_headers( + beta_headers=["compact-2026-01-12", "compact-2026-09-04"], provider="vertex_ai" + ) + + assert filtered == ["compact-2026-01-12", "compact-2026-09-04"] @pytest.mark.parametrize("provider", ["bedrock_converse", "bedrock"]) def test_fine_grained_tool_streaming_forwarded_for_bedrock(self, provider): From a084d136358a2375c0999b56afadf64ebaa4d107 Mon Sep 17 00:00:00 2001 From: tessellark <349031000+tessellark@users.noreply.github.com> Date: Fri, 30 Oct 2026 09:40:58 -0700 Subject: [PATCH 10/18] fix(passthrough): log the URL model and declared provider on Gemini and Vertex failures (#49075) * fix(passthrough): log the URL model and declared provider on Gemini and Vertex failures Pass-through failures on Gemini and Vertex routes carry the model only in the URL, so the failure spend row and the Prometheus failure counter logged an empty model. The failure payload now falls back to the model parsed from the URL on those routes, and the Prometheus provider label honors the provider the request declares before inferring one from the model name * fix(passthrough): keep model and provider on mid-stream and Vertex failure logs Build the failure payload from the URL at both streamed relay sites so a 200 stream that dies mid-body still logs the model, derive vertex_ai for Vertex routes that never declare a provider, and have the Vertex logging handler reuse get_vertex_model_id_from_url instead of its own regex * fix(passthrough): bound the Prometheus provider label and name Vertex failures like the success handler Declared providers only reach the api_provider label when they name a registered provider, otherwise the label falls back to inference. Vertex failure rows now use the success handler's model naming (vertex_ai/ prefix on rawPredict and streamRawPredict, vertex_ai/search_api on :search) and its host-based provider detection, and the detected-stream failure sites are covered by tests. * fix(passthrough): name Google failures from the URL on untracked methods too A Gemini or Vertex pass-through failure on a method the success handler does not track (countTokens, for one) still logged model "" and no provider, although the model sits in the URL. The failure payload now falls back to the Google AI host for the provider and reads /models/ from the URL, so countTokens 400s name their model and provider like generateContent ones do. Non-Google hosts keep the tracked-route rule and still log nothing extra. * fix(passthrough): trust the declared Google provider and Vertex-shaped paths on custom bases The failure path only named the Google provider from the URL when the method was one the success handler tracks or the host was a Google API host, so a Gemini countTokens failure behind a custom GEMINI_API_BASE (or a Vertex one behind a custom base) still logged an empty model. A route that declares gemini or vertex_ai now wins outright, and a URL whose path carries /projects/

/locations// counts as Vertex on any host, while a tracked method on a non-Google host no longer passes as Vertex. * fix(pass_through): stop the URL model parser at the query string so a configured key never lands in a spend row's model * test(pass_through): run the URL-model failure tests through the real ProxyLogging with a recording callback * fix(passthrough): scope declared provider to pass-through failures and match Google hosts only Prometheus trusts the resolved provider as before and reads the request-declared provider only on pass-through failures. Vertex attribution matches Google hosts, strips operation poll suffixes, prefixes vertex_ai/ only for rawPredict, reads the URL path only, covers Vertex Live WebSocket failures, and labels a router alias failure with the alias the client requested * fix(passthrough): attribute failures from the URL only on built-in Google routes and keep other routes' metric labels A user-defined pass-through route to a Google host keeps its empty model and provider, since its URL names an upstream model rather than a LiteLLM-managed one. Prometheus failure labels read the route's provider and model_group only for gemini and vertex_ai, so Azure, OpenRouter, Bedrock and other pass-through routes keep their earlier labels. The Gemini URL parser drops the query string, Google host matching needs a dot or dash boundary, and the Vertex Claude prefix follows alt=sse like the success handler. * fix(passthrough): log the URL model on guardrail blocks and the named model on Vertex Live failures A pre-call guardrail block on a pass-through route now builds its failure payload the same way upstream failures do, so the model and provider read from the URL reach spend logs and metrics. Vertex Live websocket failures (connect errors, upstream rejections, and the credential failure) now carry the model from the ?model= query parameter under the vertex_ai provider * fix(passthrough): mark guardrail-block and Vertex Live credential failures as pass-through calls Prometheus reads a failure's declared provider only on pass-through calls, so without call_type it fell back to guessing from the model name. That labeled a guardrail block on the Gemini route as vertex_ai and the Live credential failure as None * fix(passthrough): let the route name a failure's provider and give guardrail blocks their own spend row * fix(pass-through): keep URL model attribution off routes that name another provider A built-in route that names a non-Google provider, like /openai pointed at Gemini's OpenAI-compatible host, no longer takes its failure model from the Google URL. A guardrail block response now returns the x-litellm-call-id of its spend row. --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> --- litellm/integrations/prometheus.py | 53 +- litellm/llms/vertex_ai/common_utils.py | 2 +- .../llm_passthrough_endpoints.py | 12 +- .../gemini_passthrough_logging_handler.py | 8 +- .../vertex_passthrough_logging_handler.py | 63 +- .../pass_through_endpoints.py | 136 ++- .../integrations/test_prometheus_labels.py | 97 +++ .../vertex_ai/test_vertex_ai_common_utils.py | 3 + .../test_llm_pass_through_endpoints.py | 53 +- .../test_pass_through_endpoints.py | 772 +++++++++++++++--- 10 files changed, 1025 insertions(+), 174 deletions(-) diff --git a/litellm/integrations/prometheus.py b/litellm/integrations/prometheus.py index e6318525b88..5c2f0f553a8 100644 --- a/litellm/integrations/prometheus.py +++ b/litellm/integrations/prometheus.py @@ -82,6 +82,7 @@ from litellm.types.proxy.carried_budget_state import ( UserBudgetSnapshot, ) from litellm.types.utils import ( + CallTypes, StandardLoggingGuardrailInformation, StandardLoggingPayload, StandardLoggingZeroCostDiagnostic, @@ -279,6 +280,11 @@ def _logged_fields(value: object) -> Mapping[str, object]: return _NO_LOGGED_FIELDS +_GOOGLE_PASS_THROUGH_PROVIDERS: Final = frozenset( + (litellm.LlmProviders.GEMINI.value, litellm.LlmProviders.VERTEX_AI.value) +) + + def _is_llm_provider(name: str) -> bool: routable: Final = ( name in litellm.provider_list or JSONProviderRegistry.exists(name) or is_registered_custom_provider(name) @@ -2942,6 +2948,22 @@ class PrometheusLogger(CustomLogger): return False + @staticmethod + def _google_pass_through_provider(request_data: Mapping[str, object]) -> str | None: + declared_provider: Final = request_data.get("custom_llm_provider") + if request_data.get("call_type") != CallTypes.pass_through.value or not isinstance(declared_provider, str): + return None + return declared_provider if declared_provider in _GOOGLE_PASS_THROUGH_PROVIDERS else None + + @staticmethod + def _failed_request_requested_model(request_data: Mapping[str, object]) -> object: + requested_model: Final = request_data.get("model", "") + if PrometheusLogger._google_pass_through_provider(request_data) is None: + return requested_model + request_metadata: Final = _logged_fields(_logged_fields(request_data.get("litellm_params")).get("metadata")) + model_group: Final = request_metadata.get("model_group") + return model_group if isinstance(model_group, str) and model_group else requested_model + @staticmethod def _extract_api_provider_from_request_data(request_data: dict) -> str | None: """ @@ -2950,22 +2972,27 @@ class PrometheusLogger(CustomLogger): A request can fail before a deployment is resolved, so the provider is not always known. Prefer the resolved ``custom_llm_provider`` on ``litellm_params``, then any provider recovered onto a partial - ``standard_logging_object`` (e.g. a stream that broke mid-flight), and - finally infer it from the requested model name (e.g. ``gpt-4o-mini`` -> - ``openai``) since the proxy's failure ``request_data`` usually carries - only the client-supplied model. Return ``None`` when it cannot be - determined so the label emits empty rather than a guess. + ``standard_logging_object`` (e.g. a stream that broke mid-flight). On a + Gemini or Vertex pass-through failure only, next take the provider the + failure payload names (``gemini`` or ``vertex_ai``), so every other + pass-through route keeps its earlier label. Finally infer it from the + requested model name (e.g. ``gpt-4o-mini`` -> ``openai``) since the + proxy's failure ``request_data`` usually carries only the client-supplied + model. Return ``None`` when it cannot be determined so the label emits + empty rather than a guess. """ litellm_params: Final = request_data.get("litellm_params") or {} - provider = litellm_params.get("custom_llm_provider") - if provider: - return provider standard_logging_object: Final = request_data.get("standard_logging_object") or {} - provider = standard_logging_object.get("custom_llm_provider") - if provider: - return provider + resolved_provider: Final = litellm_params.get("custom_llm_provider") or standard_logging_object.get( + "custom_llm_provider" + ) + if resolved_provider: + return resolved_provider + google_provider: Final = PrometheusLogger._google_pass_through_provider(request_data) + if google_provider is not None: + return google_provider model: Final = litellm_params.get("model") or request_data.get("model") - if not model: + if not isinstance(model, str) or not model: return None try: return litellm.get_llm_provider(model=model)[1] or None @@ -3052,7 +3079,7 @@ class PrometheusLogger(CustomLogger): team_alias=user_api_key_dict.team_alias, org_id=user_api_key_dict.org_id, org_alias=user_api_key_dict.organization_alias, - requested_model=_bounded_requested_model_label(request_data.get("model", "")), + requested_model=_bounded_requested_model_label(self._failed_request_requested_model(request_data)), status_code=str(status_code), exception_status=str(status_code), exception_class=self._get_exception_class_name(original_exception), diff --git a/litellm/llms/vertex_ai/common_utils.py b/litellm/llms/vertex_ai/common_utils.py index 0fa80cb2cad..ad64d852ef9 100644 --- a/litellm/llms/vertex_ai/common_utils.py +++ b/litellm/llms/vertex_ai/common_utils.py @@ -1257,7 +1257,7 @@ def get_vertex_model_id_from_url(url: str) -> str | None: `https://${LOCATION}-aiplatform.googleapis.com/v1/projects/${PROJECT_ID}/locations/${LOCATION}/publishers/google/models/${MODEL_ID}:streamGenerateContent` """ - match: Final = re.search(r"/models/([^:]+)", url) + match: Final = re.search(r"/models/([^:?#]+)", url) return match.group(1) if match else None diff --git a/litellm/proxy/pass_through_endpoints/llm_passthrough_endpoints.py b/litellm/proxy/pass_through_endpoints/llm_passthrough_endpoints.py index 8de7441121f..33b18c3d939 100644 --- a/litellm/proxy/pass_through_endpoints/llm_passthrough_endpoints.py +++ b/litellm/proxy/pass_through_endpoints/llm_passthrough_endpoints.py @@ -32,6 +32,7 @@ from typing_extensions import ReadOnly, TypedDict import litellm from litellm import get_llm_provider from litellm._logging import verbose_proxy_logger +from litellm._uuid import uuid from litellm.constants import ( ALLOWED_VERTEX_AI_PASSTHROUGH_HEADERS, AWS_ERROR_TYPE_HEADER, @@ -149,7 +150,7 @@ from litellm.types.passthrough_endpoints.tinyfish import ( ) from litellm.types.passthrough_endpoints.vertex_ai import VertexPassThroughCredentials from litellm.types.router import LiteLLMParamsTypedDict -from litellm.types.utils import LlmProviders +from litellm.types.utils import CallTypes, LlmProviders from litellm.types.vector_stores import LiteLLM_ManagedVectorStore from litellm.utils import ProviderConfigManager @@ -4105,7 +4106,12 @@ async def vertex_ai_live_websocket_passthrough( await proxy_logging_obj.post_call_failure_hook( user_api_key_dict=user_api_key_dict, original_exception=e, - request_data={}, + request_data={ + "model": model or "", + "custom_llm_provider": litellm.LlmProviders.VERTEX_AI.value, + "call_type": CallTypes.pass_through.value, + "litellm_call_id": str(uuid.uuid4()), + }, ) if websocket.client_state != WebSocketState.DISCONNECTED: await websocket.close(code=1011, reason=VERTEX_LIVE_UNCONFIGURED_CLOSE_REASON) @@ -4140,6 +4146,8 @@ async def vertex_ai_live_websocket_passthrough( forward_headers=False, endpoint="/vertex_ai/live", accept_websocket=False, + builtin_provider_route=True, + model=model, setup_model_rewriter=_build_vertex_live_setup_model_rewriter( vertex_project=resolved_project, vertex_location=resolved_location, diff --git a/litellm/proxy/pass_through_endpoints/llm_provider_handlers/gemini_passthrough_logging_handler.py b/litellm/proxy/pass_through_endpoints/llm_provider_handlers/gemini_passthrough_logging_handler.py index 7a5d5a1e259..d449f62af2c 100644 --- a/litellm/proxy/pass_through_endpoints/llm_provider_handlers/gemini_passthrough_logging_handler.py +++ b/litellm/proxy/pass_through_endpoints/llm_provider_handlers/gemini_passthrough_logging_handler.py @@ -1,4 +1,3 @@ -import re from collections.abc import Sequence from datetime import datetime from typing import TYPE_CHECKING, Any, Final @@ -9,6 +8,7 @@ import litellm from litellm._logging import verbose_proxy_logger from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj from litellm.llms.gemini.videos.transformation import GeminiVideoConfig +from litellm.llms.vertex_ai.common_utils import get_vertex_model_id_from_url from litellm.proxy._types import PassThroughEndpointLoggingTypedDict from litellm.proxy.pass_through_endpoints.llm_provider_handlers.vertex_passthrough_logging_handler import ( GeminiSseChunkParser, @@ -207,11 +207,7 @@ class GeminiPassthroughLoggingHandler: @staticmethod def extract_model_from_url(url: str) -> str: - pattern: Final = r"/models/([^:]+)" - match: Final = re.search(pattern, url) - if match: - return match.group(1) - return "unknown" + return get_vertex_model_id_from_url(url) or "unknown" @staticmethod def _create_gemini_response_logging_payload_for_generate_content( diff --git a/litellm/proxy/pass_through_endpoints/llm_provider_handlers/vertex_passthrough_logging_handler.py b/litellm/proxy/pass_through_endpoints/llm_provider_handlers/vertex_passthrough_logging_handler.py index b3cbad91624..d5c162296d3 100644 --- a/litellm/proxy/pass_through_endpoints/llm_provider_handlers/vertex_passthrough_logging_handler.py +++ b/litellm/proxy/pass_through_endpoints/llm_provider_handlers/vertex_passthrough_logging_handler.py @@ -3,7 +3,7 @@ import re from collections.abc import Mapping, Sequence from datetime import datetime from typing import TYPE_CHECKING, Any, Final, Literal, Protocol, cast -from urllib.parse import urlparse +from urllib.parse import parse_qs, urlparse import httpx from pydantic import ConfigDict, TypeAdapter @@ -19,6 +19,7 @@ from litellm.litellm_core_utils.llm_cost_calc.usage_object_transformation import from litellm.llms.vertex_ai.common_utils import ( get_vertex_ai_lyria_generation_cost, get_vertex_location_from_url, + get_vertex_model_id_from_url, ) from litellm.llms.vertex_ai.gemini.vertex_and_google_ai_studio_gemini import ( ModelResponseIterator as VertexModelResponseIterator, @@ -47,6 +48,19 @@ from litellm.types.utils import ( ) vertex_search_api_config: Final = VertexSearchAPIVectorStoreConfig() +GEMINI_API_HOST_SUFFIX: Final = "generativelanguage.googleapis.com" +GOOGLE_AI_HOST_SUFFIXES: Final = ( + GEMINI_API_HOST_SUFFIX, + "aiplatform.googleapis.com", + "discoveryengine.googleapis.com", +) +_VERTEX_MULTI_REGION_HOST: Final = re.compile(r"aiplatform\.[a-z0-9-]+\.rep\.googleapis\.com") + + +def _is_host_or_subdomain(hostname: str, suffix: str) -> bool: + return hostname == suffix or hostname.endswith((f".{suffix}", f"-{suffix}")) + + if TYPE_CHECKING: from litellm.types.utils import LiteLLMBatch @@ -684,11 +698,18 @@ class VertexPassthroughLoggingHandler: @staticmethod def extract_model_from_url(url: str) -> str: - pattern: Final = r"/models/([^:]+)" - match: Final = re.search(pattern, url) - if match: - return match.group(1) - return "unknown" + return get_vertex_model_id_from_url(url) or "unknown" + + @staticmethod + def model_from_url_route(url_route: str) -> str | None: + parsed_url: Final = urlparse(url_route) + model: Final = get_vertex_model_id_from_url(parsed_url.path) + if model is None: + return "vertex_ai/search_api" if ":search" in parsed_url.path else None + streamed: Final = "sse" in parse_qs(parsed_url.query).get("alt", ()) + if parsed_url.path.endswith((":rawPredict", ":streamRawPredict")) and not streamed: + return f"vertex_ai/{model}" + return model @staticmethod def extract_model_name_from_vertex_path(vertex_model_path: str) -> str: @@ -734,18 +755,32 @@ class VertexPassthroughLoggingHandler: return None @staticmethod - def _get_custom_llm_provider(url: str, logging_obj: LiteLLMLoggingObj) -> str: + def is_google_ai_url(url: str) -> bool: hostname: Final = urlparse(url).hostname - if logging_obj.model_call_details.get("custom_llm_provider") == litellm.LlmProviders.GEMINI.value or ( - hostname is not None - and ( - hostname == "generativelanguage.googleapis.com" - or hostname.endswith(".generativelanguage.googleapis.com") - ) - ): + return hostname is not None and ( + any(_is_host_or_subdomain(hostname, suffix) for suffix in GOOGLE_AI_HOST_SUFFIXES) + or _VERTEX_MULTI_REGION_HOST.fullmatch(hostname) is not None + ) + + @staticmethod + def is_gemini_api_host(url: str) -> bool: + hostname: Final = urlparse(url).hostname + return hostname is not None and ( + hostname == GEMINI_API_HOST_SUFFIX or hostname.endswith(f".{GEMINI_API_HOST_SUFFIX}") + ) + + @staticmethod + def custom_llm_provider_from_url(url: str) -> str: + if VertexPassthroughLoggingHandler.is_gemini_api_host(url): return litellm.LlmProviders.GEMINI.value return litellm.LlmProviders.VERTEX_AI.value + @staticmethod + def _get_custom_llm_provider(url: str, logging_obj: LiteLLMLoggingObj) -> str: + if logging_obj.model_call_details.get("custom_llm_provider") == litellm.LlmProviders.GEMINI.value: + return litellm.LlmProviders.GEMINI.value + return VertexPassthroughLoggingHandler.custom_llm_provider_from_url(url) + @staticmethod def _is_multimodal_embedding_response(json_response: dict) -> bool: """ diff --git a/litellm/proxy/pass_through_endpoints/pass_through_endpoints.py b/litellm/proxy/pass_through_endpoints/pass_through_endpoints.py index bda2bfb9e04..cfac28db1a9 100644 --- a/litellm/proxy/pass_through_endpoints/pass_through_endpoints.py +++ b/litellm/proxy/pass_through_endpoints/pass_through_endpoints.py @@ -3,6 +3,7 @@ import asyncio import copy import json import posixpath +import re import traceback from base64 import b64encode from collections.abc import AsyncGenerator, AsyncIterator, Callable, Iterable, Mapping, Sequence @@ -77,6 +78,7 @@ from litellm.llms.base_llm.managed_resources.utils import ( from litellm.llms.custom_httpx.http_handler import get_async_httpx_client from litellm.llms.eu_jev.common_utils import eu_jev_scoped_model, eu_jev_upstream_body from litellm.llms.oss_decision import validate_oss_request +from litellm.llms.vertex_ai.common_utils import get_vertex_model_id_from_url from litellm.passthrough import BasePassthroughUtils from litellm.proxy._lazy_features import lazy_owned_routes from litellm.proxy._types import ( @@ -147,11 +149,12 @@ from litellm.types.passthrough_endpoints.pass_through_endpoints import ( EndpointType, PassthroughStandardLoggingPayload, ) -from litellm.types.utils import TRUSTED_CALLBACK_VARS_FIELD, Usage +from litellm.types.utils import TRUSTED_CALLBACK_VARS_FIELD, CallTypes, Usage from .llm_provider_handlers.tinyfish_passthrough_logging_handler import ( is_tinyfish_agent_url, ) +from .llm_provider_handlers.vertex_passthrough_logging_handler import VertexPassthroughLoggingHandler from .streaming_handler import PassThroughStreamingHandler from .success_handler import PassThroughEndpointLogging from .upstream_usage_headers import ( @@ -874,13 +877,45 @@ def _carry_guardrail_logging_info(request_data: dict, guardrail_data: dict | Non metadata.setdefault("standard_logging_guardrail_information", list(entries)) +_GOOGLE_PASSTHROUGH_PROVIDERS: Final = frozenset( + (litellm.LlmProviders.GEMINI.value, litellm.LlmProviders.VERTEX_AI.value) +) + + +def _passthrough_google_provider(url_route: str, custom_llm_provider: str | None) -> str | None: + if custom_llm_provider is not None: + return custom_llm_provider if custom_llm_provider in _GOOGLE_PASSTHROUGH_PROVIDERS else None + if VertexPassthroughLoggingHandler.is_google_ai_url(url_route): + return VertexPassthroughLoggingHandler.custom_llm_provider_from_url(url_route) + return None + + +_OPERATION_POLL_SUFFIX: Final = re.compile(r"/operations/[^/]+$") + + +def _without_operation_poll_suffix(model: str | None) -> str | None: + return _OPERATION_POLL_SUFFIX.sub("", model) if model else model + + +def _passthrough_google_model(url_route: str, google_provider: str | None) -> str | None: + match google_provider: + case "gemini": + return _without_operation_poll_suffix(get_vertex_model_id_from_url(urlparse(url_route).path)) + case "vertex_ai": + return _without_operation_poll_suffix(VertexPassthroughLoggingHandler.model_from_url_route(url_route)) + case _: + return None + + def _build_passthrough_failure_request_payload( parsed_body: dict | None, kwargs: dict | None, logging_obj: LiteLLMLoggingObj | None, custom_llm_provider: str | None, upstream_usage: UpstreamReportedUsage | None = None, + url: httpx.URL | None = None, upstream_model: str | None = None, + builtin_provider_route: bool = False, ) -> dict: """Build the ``request_data`` dict passed to ``post_call_failure_hook``. @@ -893,16 +928,25 @@ def _build_passthrough_failure_request_payload( comes paired with a usage object, so both keys are written together. ``upstream_model`` is the model name the upstream reported, which replaces the body's ``model`` on the failure row just as it does on a success row. + ``builtin_provider_route`` lets a Gemini or Vertex URL name the failure's + model and provider. A user-defined route never gets that, since its URL + names an upstream model rather than a LiteLLM-managed one. """ request_payload: Final[dict] = dict(cast(Mapping[str, object], parsed_body or {})) # cast-ok: json body if kwargs: request_payload.update(kwargs) if logging_obj is not None: request_payload["litellm_logging_obj"] = logging_obj - if "model" not in request_payload and parsed_body and isinstance(parsed_body, dict): - request_payload["model"] = parsed_body.get("model", "") - if "custom_llm_provider" not in request_payload and custom_llm_provider: - request_payload["custom_llm_provider"] = custom_llm_provider + url_route: Final = str(url) if url is not None else "" + google_provider: Final = ( + _passthrough_google_provider(url_route, custom_llm_provider) if builtin_provider_route else None + ) + if "model" not in request_payload: + request_payload["model"] = _passthrough_google_model(url_route, google_provider) or "" + resolved_provider: Final = custom_llm_provider or google_provider + request_payload.pop("custom_llm_provider", None) + if resolved_provider: + request_payload["custom_llm_provider"] = resolved_provider if upstream_usage is not None: request_payload["response_cost"] = upstream_usage.response_cost or 0.0 request_payload["combined_usage_object"] = Usage(total_tokens=upstream_usage.total_tokens or 0) @@ -1203,6 +1247,7 @@ async def pass_through_request( # Initialize variables ######################################################### litellm_call_id: Final = str(uuid.uuid4()) + builtin_provider_route: Final = not request_dispatched_to_pass_through_endpoint(request) url: httpx.URL | None = None # parsed request body @@ -1612,6 +1657,8 @@ async def pass_through_request( logging_obj=logging_obj, custom_llm_provider=custom_llm_provider, upstream_usage=upstream_usage, + url=url, + builtin_provider_route=builtin_provider_route, upstream_model=upstream_model, ), logging_obj=logging_obj, @@ -1651,6 +1698,8 @@ async def pass_through_request( kwargs=kwargs, logging_obj=logging_obj, custom_llm_provider=custom_llm_provider, + url=url, + builtin_provider_route=builtin_provider_route, ), ), managed_id_provider=_managed_id_provider, @@ -1710,6 +1759,8 @@ async def pass_through_request( logging_obj=logging_obj, custom_llm_provider=custom_llm_provider, upstream_usage=upstream_usage, + url=url, + builtin_provider_route=builtin_provider_route, upstream_model=upstream_model, ), logging_obj=logging_obj, @@ -1749,6 +1800,8 @@ async def pass_through_request( kwargs=kwargs, logging_obj=logging_obj, custom_llm_provider=custom_llm_provider, + url=url, + builtin_provider_route=builtin_provider_route, ), ), managed_id_provider=_managed_id_provider, @@ -1811,6 +1864,8 @@ async def pass_through_request( logging_obj=logging_obj, custom_llm_provider=custom_llm_provider, upstream_usage=upstream_usage, + url=url, + builtin_provider_route=builtin_provider_route, upstream_model=upstream_model, ) failure_request_payload["response_body"] = response_body @@ -1991,7 +2046,14 @@ async def pass_through_request( await proxy_logging_obj.post_call_failure_hook( user_api_key_dict=user_api_key_dict, original_exception=e, - request_data=e.request_data, + request_data=_build_passthrough_failure_request_payload( + parsed_body=e.request_data, + kwargs={"call_type": CallTypes.pass_through.value, "litellm_call_id": litellm_call_id}, + logging_obj=None, + custom_llm_provider=custom_llm_provider, + url=url, + builtin_provider_route=builtin_provider_route, + ), ) except Exception: verbose_proxy_logger.warning( @@ -2010,6 +2072,7 @@ async def pass_through_request( content=json.dumps(error_body), status_code=200, media_type="application/json", + headers={"x-litellm-call-id": litellm_call_id}, ) except Exception as e: custom_headers = ProxyBaseLLMRequestProcessing.get_custom_headers( @@ -2033,19 +2096,14 @@ async def pass_through_request( # Monitoring: Trigger post_call_failure_hook # for pass through endpoint failure ######################################################### - request_payload: Final[dict] = _parsed_body or {} - # add user_api_key_dict, litellm_call_id, passthrough_logging_payloa for logging - if kwargs: - for key, value in kwargs.items(): - request_payload[key] = value - if logging_obj is not None: - request_payload["litellm_logging_obj"] = logging_obj - - if "model" not in request_payload and _parsed_body and isinstance(_parsed_body, dict): - request_payload["model"] = _parsed_body.get("model", "") - if "custom_llm_provider" not in request_payload and custom_llm_provider: - request_payload["custom_llm_provider"] = custom_llm_provider - + request_payload: Final = _build_passthrough_failure_request_payload( + parsed_body=_parsed_body, + kwargs=kwargs, + logging_obj=logging_obj, + custom_llm_provider=custom_llm_provider, + url=url, + builtin_provider_route=builtin_provider_route, + ) _carry_guardrail_logging_info(request_payload, post_call_guardrail_data) await proxy_logging_obj.post_call_failure_hook( @@ -2547,6 +2605,8 @@ async def websocket_passthrough_request( cost_per_request: float | None = None, accept_websocket: bool = True, setup_model_rewriter: Callable[[str], str] | None = None, + builtin_provider_route: bool = False, + model: str | None = None, ): """ WebSocket passthrough request handler. @@ -2560,6 +2620,8 @@ async def websocket_passthrough_request( endpoint: The endpoint path (for logging purposes) cost_per_request: Optional field - cost per request to the target endpoint setup_model_rewriter: Optional rewrite of the setup frame's model before it reaches the upstream + builtin_provider_route: Whether a built-in provider route serves the connection, so its URL names the provider + model: The model the caller named before connecting, logged on a failure before the setup frame arrives """ from litellm.litellm_core_utils.litellm_logging import Logging from litellm.proxy.proxy_server import proxy_config, proxy_logging_obj @@ -2912,19 +2974,17 @@ async def websocket_passthrough_request( except InvalidStatus as exc: verbose_proxy_logger.exception("WebSocket passthrough (%s): upstream rejected WebSocket connection", endpoint) - # Prepare request payload for logging - request_payload = {} - if kwargs: - for key, value in kwargs.items(): - request_payload[key] = value - if logging_obj is not None: - request_payload["litellm_logging_obj"] = logging_obj - - # Log the connection failure using the same pattern as HTTP await proxy_logging_obj.post_call_failure_hook( user_api_key_dict=user_api_key_dict, original_exception=exc, - request_data=request_payload, + request_data=_build_passthrough_failure_request_payload( + parsed_body=None if model is None else {"model": model}, + kwargs=kwargs, + logging_obj=logging_obj, + custom_llm_provider=None, + url=httpx.URL(target), + builtin_provider_route=builtin_provider_route, + ), traceback_str=traceback.format_exc( limit=MAXIMUM_TRACEBACK_LINES_TO_LOG, ), @@ -2940,19 +3000,17 @@ async def websocket_passthrough_request( "WebSocket passthrough (%s): unexpected error while proxying WebSocket", endpoint ) - # Prepare request payload for logging - request_payload = {} - if kwargs: - for key, value in kwargs.items(): - request_payload[key] = value - if logging_obj is not None: - request_payload["litellm_logging_obj"] = logging_obj - - # Log the unexpected error using the same pattern as HTTP await proxy_logging_obj.post_call_failure_hook( user_api_key_dict=user_api_key_dict, original_exception=e, - request_data=request_payload, + request_data=_build_passthrough_failure_request_payload( + parsed_body=None if model is None else {"model": model}, + kwargs=kwargs, + logging_obj=logging_obj, + custom_llm_provider=None, + url=httpx.URL(target), + builtin_provider_route=builtin_provider_route, + ), traceback_str=traceback.format_exc( limit=MAXIMUM_TRACEBACK_LINES_TO_LOG, ), diff --git a/tests/unit/integrations/test_prometheus_labels.py b/tests/unit/integrations/test_prometheus_labels.py index bff89b6cbf1..878f19c66f8 100644 --- a/tests/unit/integrations/test_prometheus_labels.py +++ b/tests/unit/integrations/test_prometheus_labels.py @@ -238,6 +238,53 @@ def test_extract_api_provider_from_request_data_failure_path(): assert extract({"model": "some-unknown-model-xyz"}) is None +_PASS_THROUGH_CALL_TYPE: Final = "pass_through_endpoint" + + +@pytest.mark.parametrize( + "request_data, expected_provider", + [ + ({"model": "gpt-4o-mini", "call_type": "acompletion", "custom_llm_provider": "anthropic"}, "openai"), + ({"model": "gpt-4o-mini", "litellm_params": {"custom_llm_provider": "humanloop"}}, "humanloop"), + ({"model": "gemini-3.8-flash", "call_type": _PASS_THROUGH_CALL_TYPE}, "vertex_ai"), + ( + {"model": "gemini-3.8-flash", "call_type": _PASS_THROUGH_CALL_TYPE, "custom_llm_provider": "gemini"}, + "gemini", + ), + ( + { + "call_type": _PASS_THROUGH_CALL_TYPE, + "custom_llm_provider": "gemini", + "standard_logging_object": {"custom_llm_provider": "vertex_ai"}, + }, + "vertex_ai", + ), + ], +) +def test_extract_api_provider_reads_the_request_declared_provider_only_on_pass_through_failures( + request_data: dict[str, object], expected_provider: str +): + from litellm.integrations.prometheus import PrometheusLogger + + assert PrometheusLogger._extract_api_provider_from_request_data(request_data) == expected_provider + + +@pytest.mark.parametrize( + "declared_provider", + ["azure", "bedrock", "openrouter", "client-supplied-value", "", {"nested": "object"}, 42], +) +def test_extract_api_provider_ignores_pass_through_declared_values_outside_gemini_and_vertex( + declared_provider: object, +): + from litellm.integrations.prometheus import PrometheusLogger + + extract = PrometheusLogger._extract_api_provider_from_request_data + declared = {"call_type": _PASS_THROUGH_CALL_TYPE, "custom_llm_provider": declared_provider} + + assert extract({"model": "gpt-4o-mini", **declared}) == "openai" + assert extract({"model": "no-such-model", **declared}) is None + + def test_extract_api_provider_swallows_unknown_model_but_logs_unexpected_errors(): """ get_llm_provider raises BadRequestError for a model that maps to no @@ -849,6 +896,56 @@ async def test_failure_hook_prefers_request_data_provider_over_exception_provide ) == ["azure"] +@pytest.mark.asyncio +@pytest.mark.parametrize( + "call_type, custom_llm_provider, expected_requested_model", + [ + (_PASS_THROUGH_CALL_TYPE, "vertex_ai", "gcp/google/gemini-2.5-flash"), + ("acompletion", "vertex_ai", "other"), + (_PASS_THROUGH_CALL_TYPE, "bedrock", "other"), + ], +) +async def test_failure_hook_labels_a_pass_through_failure_with_the_router_model_the_client_requested( + call_type: str, custom_llm_provider: str, expected_requested_model: str +): + from unittest.mock import patch + + from litellm.integrations.prometheus import PrometheusLogger + from litellm.proxy._types import UserAPIKeyAuth + + router = litellm.Router( + model_list=[ + { + "model_name": "gcp/google/gemini-2.5-flash", + "litellm_params": { + "model": "vertex_ai/gemini-2.5-flash", + "vertex_project": "p", + "vertex_location": "us-central1", + }, + } + ] + ) + _clear_prometheus_registry() + try: + with patch("litellm.proxy.proxy_server.llm_router", router, create=True): # test-quality-ok: production reads proxy_server.llm_router lazily, no injection seam + await PrometheusLogger().async_post_call_failure_hook( + request_data={ + "model": "gemini-2.5-flash", + "call_type": call_type, + "custom_llm_provider": custom_llm_provider, + "metadata": {}, + "litellm_params": {"metadata": {"model_group": "gcp/google/gemini-2.5-flash"}}, + }, + original_exception=Exception("upstream 400"), + user_api_key_dict=UserAPIKeyAuth(token="tok"), + ) + assert { + s.labels.get("requested_model") for s in _collected_samples("litellm_proxy_failed_requests_metric_total") + } == {expected_requested_model} + finally: + _clear_prometheus_registry() + + def test_model_group_in_deployment_metrics(): """ Test that model_group label is present on the deployment-scoped metrics diff --git a/tests/unit/llms/vertex_ai/test_vertex_ai_common_utils.py b/tests/unit/llms/vertex_ai/test_vertex_ai_common_utils.py index 152ab2ffc13..3b222be6e5e 100644 --- a/tests/unit/llms/vertex_ai/test_vertex_ai_common_utils.py +++ b/tests/unit/llms/vertex_ai/test_vertex_ai_common_utils.py @@ -1293,6 +1293,9 @@ def test_get_vertex_model_id_from_url(): model_id = get_vertex_model_id_from_url(url) assert model_id == "gemini-pro" + url = "https://generativelanguage.googleapis.com/v1beta/models/gemini-pro?key=provider-secret#frag" + assert get_vertex_model_id_from_url(url) == "gemini-pro" + # Test with invalid URL url = "https://invalid-url.com" model_id = get_vertex_model_id_from_url(url) diff --git a/tests/unit/proxy/pass_through_endpoints/test_llm_pass_through_endpoints.py b/tests/unit/proxy/pass_through_endpoints/test_llm_pass_through_endpoints.py index aab227e9494..ff27ac54a5b 100644 --- a/tests/unit/proxy/pass_through_endpoints/test_llm_pass_through_endpoints.py +++ b/tests/unit/proxy/pass_through_endpoints/test_llm_pass_through_endpoints.py @@ -25,12 +25,14 @@ from starlette.datastructures import FormData import litellm from litellm.integrations.custom_guardrail import CustomGuardrail from litellm.integrations.custom_logger import CustomLogger +from litellm.integrations.prometheus import PrometheusLogger from tests._master_key import MASTER_KEY as SHARED_MASTER_KEY from litellm.caching.caching import DualCache -from litellm.types.utils import CallTypesLiteral +from litellm.types.utils import CallTypes, CallTypesLiteral from litellm.proxy.common_request_processing import ProxyBaseLLMRequestProcessing from litellm.exceptions import BlockedPiiEntityError, GuardrailRaisedException, ModifyResponseException from litellm.proxy._types import AwsErrorTypeProxyException, ProxyException +from litellm.proxy.spend_tracking.spend_tracking_utils import get_spend_logs_id from litellm.proxy.utils import ProxyLogging from litellm.passthrough.main import raise_for_provider_status from litellm.llms.bedrock.common_utils import BedrockError @@ -6490,6 +6492,55 @@ class TestVertexAILiveWebsocketPassthrough: assert "default_vertex_config" in close_kwargs["reason"] assert len(close_kwargs["reason"].encode("utf-8")) <= 123 + @pytest.mark.asyncio + @pytest.mark.parametrize("model, expected_model", [("gemini-3.8-live", "gemini-3.8-live"), (None, "")]) + async def test_credential_failure_logs_the_named_model_under_vertex_ai( + self, monkeypatch: pytest.MonkeyPatch, model: str | None, expected_model: str + ): + from litellm.proxy.pass_through_endpoints import ( + llm_passthrough_endpoints as passthrough_module, + ) + + class _FailureRecorder(CustomLogger): + def __init__(self) -> None: + super().__init__() + self.request_data: tuple[dict[str, object], ...] = () + + async def async_post_call_failure_hook( + self, + request_data: dict[str, object], + original_exception: Exception, + user_api_key_dict: UserAPIKeyAuth, + traceback_str: str | None = None, + ) -> None: + self.request_data = (*self.request_data, request_data) + + recorder: Final = _FailureRecorder() + monkeypatch.setattr("litellm.proxy.proxy_server.llm_router", None) + monkeypatch.setattr( + "litellm.proxy.proxy_server.proxy_logging_obj", ProxyLogging(user_api_key_cache=DualCache()) + ) + monkeypatch.setattr(litellm, "callbacks", [recorder]) + monkeypatch.setattr(passthrough_module.passthrough_endpoint_router, "default_vertex_config", None) + self._clear_vertex_env(monkeypatch) + ensure_token: Final = AsyncMock(side_effect=Exception("Unable to find your credentials")) + + with patch.object(passthrough_module.vertex_llm_base, "ensure_access_token_async", ensure_token): + await passthrough_module.vertex_ai_live_websocket_passthrough( + websocket=self._websocket(), + model=model, + user_api_key_dict=UserAPIKeyAuth(), + ) + + assert [ + ( + d["model"], + PrometheusLogger._extract_api_provider_from_request_data(d), + get_spend_logs_id(CallTypes.pass_through.value, {}, d) is not None, + ) + for d in recorder.request_data + ] == [(expected_model, "vertex_ai", True)] + class TestAnthropicProxyRoute: """The /anthropic passthrough route: custom auth headers must not clobber the diff --git a/tests/unit/proxy/pass_through_endpoints/test_pass_through_endpoints.py b/tests/unit/proxy/pass_through_endpoints/test_pass_through_endpoints.py index 7b62853b0e9..a9098ac3d59 100644 --- a/tests/unit/proxy/pass_through_endpoints/test_pass_through_endpoints.py +++ b/tests/unit/proxy/pass_through_endpoints/test_pass_through_endpoints.py @@ -30,20 +30,26 @@ from starlette.datastructures import UploadFile as StarletteUploadFile import litellm from litellm._logging import verbose_proxy_logger +from litellm.caching.dual_cache import DualCache from litellm.constants import DEFAULT_REQUEST_TIMEOUT_SECONDS from litellm.integrations.custom_guardrail import CustomGuardrail from litellm.integrations.custom_logger import CustomLogger +from litellm.integrations.prometheus import PrometheusLogger from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj from litellm.proxy._lazy_features import LazyFeature, attach_lazy_features from litellm.proxy._types import ProxyException, UserAPIKeyAuth from litellm.proxy.litellm_pre_call_utils import LiteLLMProxyRequestSetup from litellm.proxy.pass_through_endpoints.llm_passthrough_endpoints import router as llm_passthrough_router +from litellm.proxy.pass_through_endpoints.llm_provider_handlers.vertex_passthrough_logging_handler import ( + VertexPassthroughLoggingHandler, +) from litellm.proxy.pass_through_endpoints.pass_through_endpoints import ( DEFAULT_PASS_THROUGH_REQUEST_TIMEOUT_SECONDS, LITELLM_PASS_THROUGH_CUSTOM_BODY_STATE_KEY, HttpPassThroughEndpointHelpers, InitPassThroughEndpointHelpers, SafeRouteAdder, + _build_passthrough_failure_request_payload, _registered_pass_through_routes, _truncate_upstream_error_body, _update_metadata_with_tags_in_header, @@ -61,8 +67,11 @@ from litellm.proxy.pass_through_endpoints.success_handler import ( PassThroughEndpointLogging, ) from litellm.proxy.route_llm_request import ProxyModelNotFoundError +from litellm.proxy.spend_tracking.spend_tracking_utils import get_spend_logs_id +from litellm.proxy.utils import ProxyLogging from litellm.types import utils as types_utils from litellm.types.guardrails import GuardrailEventHooks +from litellm.types.utils import CallTypes from litellm.types.passthrough_endpoints.pass_through_endpoints import ( LITELLM_PASS_THROUGH_DEPLOYMENT_MODEL_INFO_STATE_KEY, LITELLM_PASS_THROUGH_RAW_BODY_STATE_KEY, @@ -140,9 +149,7 @@ def test_is_multipart(): ), ], ) -def test_get_endpoint_type_classifies_anthropic_messages_routes( - url: str, expected_endpoint_type: EndpointType -) -> None: +def test_get_endpoint_type_classifies_anthropic_messages_routes(url: str, expected_endpoint_type: EndpointType) -> None: assert HttpPassThroughEndpointHelpers.get_endpoint_type(url) == expected_endpoint_type @@ -430,53 +437,18 @@ async def test_pass_through_request_failure_handler(): Critical Test: When a users pass through endpoint request fails, we must log the failure code, exception in litellm spend logs. """ - with patch("litellm.proxy.proxy_server.proxy_logging_obj") as mock_proxy_logging: - with patch("litellm.llms.custom_httpx.http_handler.get_async_httpx_client") as mock_get_client: - with patch( - "litellm.proxy.pass_through_endpoints.pass_through_endpoints.ProxyBaseLLMRequestProcessing" - ) as mock_processing: - # Setup mock for post_call_failure_hook and pre_call_hook - mock_proxy_logging.post_call_failure_hook = AsyncMock() - mock_proxy_logging.pre_call_hook = AsyncMock() + request_failure = httpx.HTTPError("Request failed") + user_api_key_dict = UserAPIKeyAuth(api_key="sk-test") - # Setup mock for httpx client - mock_client = MagicMock() - mock_client.client = MagicMock() - mock_client.client.request = AsyncMock(side_effect=httpx.HTTPError("Request failed")) - mock_get_client.return_value = mock_client + failure: Final = await _run_passthrough_failure( + "http://test.com", None, request_failure, user_api_key_dict=user_api_key_dict + ) - # Mock headers for custom headers - mock_processing.get_custom_headers.return_value = {} - - # Create mock request - mock_request = MagicMock(spec=Request) - mock_request.method = "POST" - mock_request.body = AsyncMock(return_value=b'{"test": "data"}') - mock_request.headers = Headers({}) - - # Create a simple empty QueryParams - mock_request.query_params = QueryParams({}) - - # Create mock user API key dict - mock_user_api_key_dict = MagicMock() - - # Call the function with a target that will trigger an HTTPError - with pytest.raises(ProxyException): - await pass_through_request( - request=mock_request, - target="http://test.com", - custom_headers={}, - user_api_key_dict=mock_user_api_key_dict, - ) - - # Assert post_call_failure_hook was called - mock_proxy_logging.post_call_failure_hook.assert_called_once() - - # Verify the arguments to post_call_failure_hook - call_args = mock_proxy_logging.post_call_failure_hook.call_args[1] - assert call_args["user_api_key_dict"] == mock_user_api_key_dict - assert isinstance(call_args["original_exception"], TypeError) # Now expecting TypeError - assert "traceback_str" in call_args + assert failure.user_api_key_dict is user_api_key_dict + assert failure.original_exception is request_failure + assert "Request failed" in (failure.traceback_str or "") + assert failure.request_data["contents"] == "not a list" + assert failure.request_data["model"] == "" @pytest.mark.asyncio @@ -3656,9 +3628,7 @@ async def test_pass_through_request_preserves_target_query_without_client_query( @pytest.mark.asyncio -async def test_config_pass_through_without_configured_headers_preserves_no_authorization( - tmp_path, monkeypatch -): +async def test_config_pass_through_without_configured_headers_preserves_no_authorization(tmp_path, monkeypatch): proxy = await _boot_db_backed_proxy( tmp_path, monkeypatch, @@ -3725,9 +3695,7 @@ async def test_pass_through_rpm_limit_is_sequential_and_scoped_per_key(tmp_path, proxy = await _boot_db_backed_proxy( tmp_path, monkeypatch, - config_pass_through_endpoints=[ - {"path": "/rpm-rerank", "target": "http://upstream.test/rerank", "auth": True} - ], + config_pass_through_endpoints=[{"path": "/rpm-rerank", "target": "http://upstream.test/rerank", "auth": True}], db_pass_through_endpoints=[], master_key=MASTER_KEY, ) @@ -4902,8 +4870,10 @@ from fastapi import HTTPException as _FastAPIHTTPException from litellm.exceptions import ( BlockedPiiEntityError, GuardrailRaisedException, + ModifyResponseException, ) from litellm.litellm_core_utils.logging_worker import GLOBAL_LOGGING_WORKER + _PT_MODULE = "litellm.proxy.pass_through_endpoints.pass_through_endpoints" @@ -8182,8 +8152,10 @@ async def test_user_defined_passthrough_is_neither_tracked_nor_enforced(metadata } ) body: Final = { - "model": "upstream-only-model", metadata_slot: { - "model_group": "managed-model", "customer_label": "retained", + "model": "upstream-only-model", + metadata_slot: { + "model_group": "managed-model", + "customer_label": "retained", "user_api_key_team_model_max_budget": budget, }, } @@ -8191,48 +8163,84 @@ async def test_user_defined_passthrough_is_neither_tracked_nor_enforced(metadata assert await limiter.is_team_within_model_budget("shared-team", budget, None, "managed-model") start: Final = datetime.now() logging_obj: Final = LiteLLMLoggingObj( - model="upstream-only-model", messages=[], stream=False, call_type="pass_through_endpoint", - start_time=start, litellm_call_id="custom-budget", function_id="custom-budget", kwargs={}, + model="upstream-only-model", + messages=[], + stream=False, + call_type="pass_through_endpoint", + start_time=start, + litellm_call_id="custom-budget", + function_id="custom-budget", + kwargs={}, dynamic_async_success_callbacks=[limiter], ) payload: Final = { - "url": "https://upstream.test/echo", "request_body": body, "request_method": "POST", "cost_per_request": 0.25, + "url": "https://upstream.test/echo", + "request_body": body, + "request_method": "POST", + "cost_per_request": 0.25, } kwargs: Final = HttpPassThroughEndpointHelpers._init_kwargs_for_pass_through_endpoint( - request=request, user_api_key_dict=auth, passthrough_logging_payload=payload, logging_obj=logging_obj, - _parsed_body=body, litellm_call_id="custom-budget", + request=request, + user_api_key_dict=auth, + passthrough_logging_payload=payload, + logging_obj=logging_obj, + _parsed_body=body, + litellm_call_id="custom-budget", ) logging_obj.update_environment_variables( - model="upstream-only-model", user="unknown", optional_params={}, - litellm_params=kwargs["litellm_params"], call_type="pass_through_endpoint", + model="upstream-only-model", + user="unknown", + optional_params={}, + litellm_params=kwargs["litellm_params"], + call_type="pass_through_endpoint", ) response: Final = httpx.Response( - 200, request=httpx.Request("POST", "https://upstream.test/echo"), json={"ok": True}, + 200, + request=httpx.Request("POST", "https://upstream.test/echo"), + json={"ok": True}, ) await PassThroughEndpointLogging().pass_through_async_success_handler( - httpx_response=response, response_body={"ok": True}, request_body=body, logging_obj=logging_obj, - url_route="https://upstream.test/echo", result=response.text, start_time=start, end_time=datetime.now(), - cache_hit=False, **kwargs, + httpx_response=response, + response_body={"ok": True}, + request_body=body, + logging_obj=logging_obj, + url_route="https://upstream.test/echo", + result=response.text, + start_time=start, + end_time=datetime.now(), + cache_hit=False, + **kwargs, ) assert logging_obj.model_call_details["response_cost"] == 0.25 assert await limiter.is_team_within_model_budget("shared-team", budget, None, "managed-model") metadata: Final = kwargs["litellm_params"]["metadata"] assert (metadata["model_group"], metadata["customer_label"]) == ("managed-model", "retained") - assert metadata.keys().isdisjoint({ - "user_api_key_model_max_budget", "user_api_key_team_model_max_budget", - "user_api_key_user_model_max_budget", "user_api_key_end_user_model_max_budget", - }) + assert metadata.keys().isdisjoint( + { + "user_api_key_model_max_budget", + "user_api_key_team_model_max_budget", + "user_api_key_user_model_max_budget", + "user_api_key_end_user_model_max_budget", + } + ) @pytest.mark.parametrize("metadata_slot", ["metadata", "litellm_metadata"]) def test_builtin_passthrough_pins_model_group_to_the_resolved_model(metadata_slot: str) -> None: - request: Final = Request({ - "type": "http", "method": "POST", "path": "/gemini/v1beta/models/gemini-2.5-flash:generateContent", - "headers": [], "query_string": b"", - }) + request: Final = Request( + { + "type": "http", + "method": "POST", + "path": "/gemini/v1beta/models/gemini-2.5-flash:generateContent", + "headers": [], + "query_string": b"", + } + ) kwargs: Final = HttpPassThroughEndpointHelpers._init_kwargs_for_pass_through_endpoint( - request=request, user_api_key_dict=UserAPIKeyAuth(token="hash", user_id="u-1"), - passthrough_logging_payload=MagicMock(), logging_obj=MagicMock(), + request=request, + user_api_key_dict=UserAPIKeyAuth(token="hash", user_id="u-1"), + passthrough_logging_payload=MagicMock(), + logging_obj=MagicMock(), _parsed_body={"contents": [], metadata_slot: {"model_group": "unbounded-client-choice"}}, ) assert kwargs["litellm_params"]["metadata"]["model_group"] == "gemini-2.5-flash" @@ -8591,6 +8599,581 @@ def test_passthrough_client_cannot_forge_session_id_omission(client_metadata_key ) +_GEMINI_FLASH_TARGET: Final = "https://generativelanguage.googleapis.com/v1beta/models/gemini-3.8-flash:generateContent" +_VERTEX_FLASH_TARGET: Final = ( + "https://us-east5-aiplatform.googleapis.com/v1/projects/p/locations/us-east5" + "/publishers/google/models/gemini-3.8-flash:generateContent" +) +_VERTEX_CLAUDE_RAW_PREDICT_TARGET: Final = ( + "https://us-east5-aiplatform.googleapis.com/v1/projects/p/locations/us-east5" + "/publishers/anthropic/models/claude-opus-5:rawPredict" +) +_VERTEX_SEARCH_TARGET: Final = ( + "https://discoveryengine.googleapis.com/v1/projects/p/locations/global" + "/dataStores/default/servingConfigs/default:search" +) +_RELAYED_GEMINI_COUNT_TOKENS_TARGET: Final = "http://127.0.0.1:30732/v1beta/models/gemini-3.8-flash:countTokens" +_TRANSLATE_TARGET: Final = "https://translate.googleapis.com/v3/projects/p/locations/global:translateText" +_NON_GOOGLE_VERTEX_SHAPED_TARGET: Final = "https://llm.example.com/v1/projects/a/locations/b/models/my-model:predict" +_VERTEX_LIVE_TARGET: Final = ( + "wss://us-central1-aiplatform.googleapis.com/ws/google.cloud.aiplatform.v1.LlmBidiService/BidiGenerateContent" +) +_VEO_OPERATION_SUFFIX: Final = "/models/veo-3.0-generate-001/operations/op-8f2c1" +_RELAYED_VERTEX_COUNT_TOKENS_TARGET: Final = ( + "http://127.0.0.1:30732/v1/projects/p/locations/us-east5/publishers/google/models/gemini-3.8-flash:countTokens" +) + + +@dataclass(frozen=True, slots=True) +class _RecordedFailure: + request_data: dict[str, object] + original_exception: Exception + user_api_key_dict: UserAPIKeyAuth + traceback_str: str | None + + +class _FailureHookRecorder(CustomLogger): + def __init__(self) -> None: + super().__init__() + self.failures: tuple[_RecordedFailure, ...] = () + + async def async_post_call_failure_hook( + self, + request_data: dict[str, object], + original_exception: Exception, + user_api_key_dict: UserAPIKeyAuth, + traceback_str: str | None = None, + ) -> None: + self.failures = ( + *self.failures, + _RecordedFailure(request_data, original_exception, user_api_key_dict, traceback_str), + ) + + +@dataclass(frozen=True, slots=True) +class _PassthroughRun: + failure: _RecordedFailure + response: Response | None + + +async def _drained_pass_through_response( + request: Request, + target: str, + custom_llm_provider: str | None, + user_api_key_dict: UserAPIKeyAuth, + stream: bool, +) -> Response | None: + try: + received: Final = await pass_through_request( + request=request, + target=target, + custom_headers={}, + user_api_key_dict=user_api_key_dict, + custom_llm_provider=custom_llm_provider, + stream=stream, + ) + if isinstance(received, StreamingResponse): + async for _ in received.body_iterator: + pass + except (ProxyException, httpx.HTTPError): + return None + return received if isinstance(received, Response) else None + + +async def _run_passthrough_failure( + target: str, + custom_llm_provider: str | None, + upstream: httpx.Response | Exception, + stream: bool = False, + user_api_key_dict: UserAPIKeyAuth | None = None, + dispatched_endpoint: Callable[..., object] | None = None, + guardrail: CustomGuardrail | None = None, + body: bytes = b'{"contents": "not a list"}', +) -> _RecordedFailure: + run: Final = await _run_passthrough( + target, custom_llm_provider, upstream, stream, user_api_key_dict, dispatched_endpoint, guardrail, body + ) + return run.failure + + +async def _run_passthrough( + target: str, + custom_llm_provider: str | None, + upstream: httpx.Response | Exception, + stream: bool = False, + user_api_key_dict: UserAPIKeyAuth | None = None, + dispatched_endpoint: Callable[..., object] | None = None, + guardrail: CustomGuardrail | None = None, + body: bytes = b'{"contents": "not a list"}', +) -> _PassthroughRun: + recorder: Final = _FailureHookRecorder() + with ( + pytest.MonkeyPatch.context() as litellm_patch, + patch( # test-quality-ok: proxy_logging_obj is a proxy_server module global read inside pass_through_request; there is no injection seam + "litellm.proxy.proxy_server.proxy_logging_obj", ProxyLogging(user_api_key_cache=DualCache()) + ), + respx.mock(assert_all_called=False) as upstream_mock, + ): + litellm_patch.setattr(litellm, "disable_aiohttp_transport", True) + litellm_patch.setattr(litellm, "callbacks", [recorder] if guardrail is None else [guardrail, recorder]) + litellm.in_memory_llm_clients_cache.flush_cache() + upstream_route: Final = upstream_mock.route(method="POST") + if isinstance(upstream, Exception): + upstream_route.mock(side_effect=upstream) + else: + upstream_route.mock(return_value=upstream) + mock_request: Final = MagicMock(spec=Request) + mock_request.method = "POST" + proxy_url: Final = httpx.URL(target.replace("https://", "http://test-proxy.com/")) + mock_request.url = proxy_url + mock_request.scope = { + "type": "http", + "method": "POST", + "path": proxy_url.path, + **({"endpoint": dispatched_endpoint} if dispatched_endpoint is not None else {}), + } + mock_request.body = AsyncMock(return_value=body) + mock_request.headers = Headers({"content-type": "application/json"}) + mock_request.query_params = QueryParams({}) + response: Final = await _drained_pass_through_response( + mock_request, + target, + custom_llm_provider, + user_api_key_dict or UserAPIKeyAuth(api_key="sk-test"), + stream, + ) + await asyncio.sleep(0) + litellm.in_memory_llm_clients_cache.flush_cache() + assert len(recorder.failures) == 1 + return _PassthroughRun(recorder.failures[0], response) + + +def _upstream_400(target: str) -> httpx.Response: + return httpx.Response( + status_code=400, + headers={"content-type": "application/json"}, + content=json.dumps({"error": {"code": 400, "status": "INVALID_ARGUMENT"}}).encode(), + request=httpx.Request("POST", target), + ) + + +def _upstream_400_as_event_stream(target: str) -> httpx.Response: + return httpx.Response( + status_code=400, + headers={"content-type": "text/event-stream"}, + content=b'data: {"error": {"code": 400, "status": "INVALID_ARGUMENT"}}\n\n', + request=httpx.Request("POST", target), + ) + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "target, custom_llm_provider, stream, expected_provider", + [ + (_GEMINI_FLASH_TARGET, "gemini", False, "gemini"), + (_GEMINI_FLASH_TARGET, "gemini", True, "gemini"), + (_VERTEX_FLASH_TARGET, None, False, "vertex_ai"), + (_VERTEX_FLASH_TARGET, None, True, "vertex_ai"), + (_GEMINI_FLASH_TARGET.replace(":generateContent", ":countTokens"), "gemini", False, "gemini"), + (_VERTEX_FLASH_TARGET.replace(":generateContent", ":countTokens"), None, False, "vertex_ai"), + (_RELAYED_GEMINI_COUNT_TOKENS_TARGET, "gemini", False, "gemini"), + ], +) +async def test_passthrough_upstream_error_logs_model_from_url( + target: str, custom_llm_provider: str | None, stream: bool, expected_provider: str +): + failure: Final = await _run_passthrough_failure(target, custom_llm_provider, _upstream_400(target), stream=stream) + assert failure.request_data["model"] == "gemini-3.8-flash" + assert failure.request_data["custom_llm_provider"] == expected_provider + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "target", + [ + _RELAYED_VERTEX_COUNT_TOKENS_TARGET, + _TRANSLATE_TARGET, + _NON_GOOGLE_VERTEX_SHAPED_TARGET, + ], +) +async def test_passthrough_upstream_error_to_a_host_outside_gemini_and_vertex_gets_no_url_attribution(target: str): + failure: Final = await _run_passthrough_failure(target, None, _upstream_400(target)) + assert failure.request_data["model"] == "" + assert failure.request_data.get("custom_llm_provider") is None + + +@pytest.mark.asyncio +@pytest.mark.parametrize("target", [_GEMINI_FLASH_TARGET, _VERTEX_FLASH_TARGET]) +async def test_passthrough_upstream_error_on_a_user_defined_route_to_google_gets_no_url_attribution(target: str): + failure: Final = await _run_passthrough_failure( + target, None, _upstream_400(target), dispatched_endpoint=_marked_pass_through_endpoint() + ) + assert failure.request_data["model"] == "" + assert failure.request_data.get("custom_llm_provider") is None + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "target, custom_llm_provider, expected_provider", + [ + (_GEMINI_FLASH_TARGET, "gemini", "gemini"), + (_VERTEX_FLASH_TARGET, None, "vertex_ai"), + ], +) +async def test_passthrough_detected_stream_upstream_error_logs_model_from_url( + target: str, custom_llm_provider: str | None, expected_provider: str +): + failure: Final = await _run_passthrough_failure( + target, custom_llm_provider, _upstream_400_as_event_stream(target), stream=False + ) + assert failure.request_data["model"] == "gemini-3.8-flash" + assert failure.request_data["custom_llm_provider"] == expected_provider + + +@pytest.mark.asyncio +async def test_passthrough_internal_failure_logs_model_from_url(): + failure: Final = await _run_passthrough_failure( + _GEMINI_FLASH_TARGET, "gemini", httpx.ReadTimeout("upstream timed out") + ) + assert failure.request_data["model"] == "gemini-3.8-flash" + assert failure.request_data["custom_llm_provider"] == "gemini" + + +def _upstream_dropping_stream(target: str) -> httpx.Response: + return httpx.Response( + status_code=200, + headers={"content-type": "text/event-stream"}, + stream=_UpstreamDroppingMidStream(), + request=httpx.Request("POST", target), + ) + + +@pytest.mark.asyncio +@pytest.mark.parametrize("stream", [True, False]) +@pytest.mark.parametrize( + "target, custom_llm_provider, expected_provider", + [ + (_GEMINI_FLASH_TARGET.replace(":generateContent", ":streamGenerateContent?alt=sse"), "gemini", "gemini"), + (_VERTEX_FLASH_TARGET.replace(":generateContent", ":streamGenerateContent?alt=sse"), None, "vertex_ai"), + ], +) +async def test_passthrough_mid_stream_drop_logs_model_from_url( + target: str, custom_llm_provider: str | None, expected_provider: str, stream: bool +): + failure: Final = await _run_passthrough_failure( + target, custom_llm_provider, _upstream_dropping_stream(target), stream=stream + ) + assert isinstance(failure.original_exception, httpx.ReadError) + assert failure.request_data["model"] == "gemini-3.8-flash" + assert failure.request_data["custom_llm_provider"] == expected_provider + + +class _ModifyResponseOnPreCallGuardrail(CustomGuardrail): + async def async_pre_call_hook( + self, + user_api_key_dict: UserAPIKeyAuth, + cache: DualCache, + data: dict[str, object], + call_type: str, + ) -> None: + self.raise_passthrough_exception(violation_message="blocked by policy", request_data=data, detection_info=None) + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "target, custom_llm_provider, body, expected_model, expected_provider", + [ + (_GEMINI_FLASH_TARGET, "gemini", b'{"contents": "not a list"}', "gemini-3.8-flash", "gemini"), + (_VERTEX_FLASH_TARGET, None, b'{"contents": "not a list"}', "gemini-3.8-flash", "vertex_ai"), + (_VERTEX_FLASH_TARGET, None, b'{"custom_llm_provider": "gemini"}', "gemini-3.8-flash", "vertex_ai"), + ], +) +async def test_passthrough_guardrail_block_logs_model_from_url( + target: str, custom_llm_provider: str | None, body: bytes, expected_model: str, expected_provider: str +): + guardrail: Final = _ModifyResponseOnPreCallGuardrail( + guardrail_name="block-policy", event_hook=GuardrailEventHooks.pre_call, default_on=True + ) + failure: Final = await _run_passthrough_failure( + target, custom_llm_provider, _upstream_400(target), guardrail=guardrail, body=body + ) + assert isinstance(failure.original_exception, ModifyResponseException) + assert failure.request_data["model"] == expected_model + assert failure.request_data["custom_llm_provider"] == expected_provider + assert PrometheusLogger._extract_api_provider_from_request_data(failure.request_data) == expected_provider + assert get_spend_logs_id(CallTypes.pass_through.value, {}, failure.request_data) is not None + + +@pytest.mark.asyncio +@pytest.mark.parametrize("body", [b'{"contents": "not a list"}', b'{"litellm_call_id": "client-chosen-id"}']) +async def test_passthrough_guardrail_block_response_names_the_call_id_of_its_spend_row(body: bytes): + guardrail: Final = _ModifyResponseOnPreCallGuardrail( + guardrail_name="block-policy", event_hook=GuardrailEventHooks.pre_call, default_on=True + ) + run: Final = await _run_passthrough( + _GEMINI_FLASH_TARGET, "gemini", _upstream_400(_GEMINI_FLASH_TARGET), guardrail=guardrail, body=body + ) + assert isinstance(run.failure.original_exception, ModifyResponseException) + assert run.response is not None + assert run.response.headers["x-litellm-call-id"] == get_spend_logs_id( + CallTypes.pass_through.value, {}, run.failure.request_data + ) + + +@pytest.mark.asyncio +@pytest.mark.parametrize("body", [b'{"contents": "not a list"}', b'{"custom_llm_provider": "vertex_ai"}']) +async def test_passthrough_guardrail_block_on_a_user_defined_route_gets_no_url_attribution(body: bytes): + guardrail: Final = _ModifyResponseOnPreCallGuardrail( + guardrail_name="block-policy", event_hook=GuardrailEventHooks.pre_call, default_on=True + ) + failure: Final = await _run_passthrough_failure( + _GEMINI_FLASH_TARGET, + None, + _upstream_400(_GEMINI_FLASH_TARGET), + dispatched_endpoint=_marked_pass_through_endpoint(), + guardrail=guardrail, + body=body, + ) + assert isinstance(failure.original_exception, ModifyResponseException) + assert failure.request_data["model"] == "" + assert "custom_llm_provider" not in failure.request_data + + +def _websocket_connect_failure(kind: str) -> Exception: + from websockets.datastructures import Headers as WebSocketHeaders + from websockets.exceptions import InvalidStatus + from websockets.http11 import Response as WebSocketHandshakeResponse + + if kind == "rejected_handshake": + return InvalidStatus(WebSocketHandshakeResponse(403, "Forbidden", WebSocketHeaders())) + return OSError("connection refused") + + +@pytest.mark.asyncio +@pytest.mark.parametrize("failure_kind", ["rejected_handshake", "unreachable_host"]) +@pytest.mark.parametrize( + "target, builtin_provider_route, expected_provider", + [ + (_VERTEX_LIVE_TARGET, True, "vertex_ai"), + (_VERTEX_LIVE_TARGET, False, None), + ("wss://api.openai.com/v1/realtime?model=gpt-realtime", True, None), + ], +) +async def test_websocket_passthrough_connect_failure_logs_the_provider_of_the_upstream_url( + failure_kind: str, target: str, builtin_provider_route: bool, expected_provider: str | None +): + failure: Final = await _run_websocket_connect_failure(failure_kind, target, builtin_provider_route) + assert failure.request_data.get("custom_llm_provider") == expected_provider + + +@pytest.mark.asyncio +@pytest.mark.parametrize("failure_kind", ["rejected_handshake", "unreachable_host"]) +@pytest.mark.parametrize("model, expected_model", [("gemini-3.8-live", "gemini-3.8-live"), (None, "")]) +async def test_websocket_passthrough_connect_failure_logs_the_model_the_caller_named( + failure_kind: str, model: str | None, expected_model: str +): + failure: Final = await _run_websocket_connect_failure(failure_kind, _VERTEX_LIVE_TARGET, True, model=model) + assert failure.request_data["model"] == expected_model + + +async def _run_websocket_connect_failure( + failure_kind: str, target: str, builtin_provider_route: bool, model: str | None = None +) -> _RecordedFailure: + recorder: Final = _FailureHookRecorder() + with ( + pytest.MonkeyPatch.context() as litellm_patch, + patch( # test-quality-ok: proxy_logging_obj is a proxy_server module global read inside websocket_passthrough_request; there is no injection seam + "litellm.proxy.proxy_server.proxy_logging_obj", ProxyLogging(user_api_key_cache=DualCache()) + ), + patch( + "litellm.proxy.pass_through_endpoints.pass_through_endpoints.connect", + side_effect=_websocket_connect_failure(failure_kind), + ), + ): + litellm_patch.setattr(litellm, "callbacks", [recorder]) + await websocket_passthrough_request( + websocket=_client_websocket(_pending_receive), + target=target, + custom_headers={}, + user_api_key_dict=UserAPIKeyAuth(api_key="sk-test"), + endpoint="/vertex_ai/live", + accept_websocket=False, + builtin_provider_route=builtin_provider_route, + model=model, + ) + assert len(recorder.failures) == 1 + return recorder.failures[0] + + +@pytest.mark.parametrize( + "parsed_body, url, custom_llm_provider, expected_model, expected_provider", + [ + ({"model": "from-body"}, _GEMINI_FLASH_TARGET, "gemini", "from-body", "gemini"), + ({"contents": []}, _GEMINI_FLASH_TARGET, "gemini", "gemini-3.8-flash", "gemini"), + ({"contents": []}, _VERTEX_FLASH_TARGET, None, "gemini-3.8-flash", "vertex_ai"), + ({"contents": []}, _VERTEX_FLASH_TARGET, "anthropic", "", "anthropic"), + ({}, "https://generativelanguage.googleapis.com/v1beta/openai/models/gemini-3.8-flash", "openai", "", "openai"), + ({"custom_llm_provider": "from-body"}, _VERTEX_FLASH_TARGET, None, "gemini-3.8-flash", "vertex_ai"), + ({"custom_llm_provider": "gemini"}, _VERTEX_FLASH_TARGET, "anthropic", "", "anthropic"), + ({"contents": []}, _GEMINI_FLASH_TARGET, None, "gemini-3.8-flash", "gemini"), + ( + {}, + "https://generativelanguage.googleapis.com/v1beta/models/gemini-3.8-flash?key=provider-secret", + "gemini", + "gemini-3.8-flash", + "gemini", + ), + ( + {}, + "https://us-east5-aiplatform.googleapis.com/v1/projects/p/locations/us-east5" + "/publishers/google/models/gemini-3.8-flash?key=provider-secret", + None, + "gemini-3.8-flash", + "vertex_ai", + ), + ({}, _VERTEX_CLAUDE_RAW_PREDICT_TARGET, None, "vertex_ai/claude-opus-5", "vertex_ai"), + ( + {}, + _VERTEX_CLAUDE_RAW_PREDICT_TARGET.replace(":rawPredict", ":streamRawPredict?alt=sse"), + None, + "claude-opus-5", + "vertex_ai", + ), + ({}, "https://generativelanguage.googleapis.com/v1beta/models?pageToken=/models/x", "gemini", "", "gemini"), + ( + {"contents": []}, + "https://evilaiplatform.googleapis.com/v1/projects/p/locations/l" + "/publishers/google/models/gemini-3.8-flash:generateContent", + None, + "", + None, + ), + ({}, _VERTEX_SEARCH_TARGET, None, "vertex_ai/search_api", "vertex_ai"), + ({}, f"{_VERTEX_FLASH_TARGET}?tag=rawPredict", None, "gemini-3.8-flash", "vertex_ai"), + ( + {}, + "https://us-east5-aiplatform.googleapis.com/v1/projects/p/locations/us-east5/batchPredictionJobs" + "?filter=:search", + None, + "", + "vertex_ai", + ), + ( + {}, + f"https://generativelanguage.googleapis.com/v1beta{_VEO_OPERATION_SUFFIX}", + "gemini", + "veo-3.0-generate-001", + "gemini", + ), + ( + {}, + "https://us-central1-aiplatform.googleapis.com/v1/projects/p/locations/us-central1" + f"/publishers/google{_VEO_OPERATION_SUFFIX}", + None, + "veo-3.0-generate-001", + "vertex_ai", + ), + ( + {"contents": []}, + "https://aiplatform.us.rep.googleapis.com/v1/projects/p/locations/us" + "/publishers/google/models/gemini-3.8-flash:generateContent", + None, + "gemini-3.8-flash", + "vertex_ai", + ), + ( + {}, + "https://us-east5-aiplatform.googleapis.com/v1/projects/p/locations/us-east5/batchPredictionJobs", + None, + "", + "vertex_ai", + ), + ( + {"contents": []}, + _GEMINI_FLASH_TARGET.replace(":generateContent", ":countTokens"), + "gemini", + "gemini-3.8-flash", + "gemini", + ), + ( + {"contents": []}, + _VERTEX_FLASH_TARGET.replace(":generateContent", ":countTokens"), + None, + "gemini-3.8-flash", + "vertex_ai", + ), + ( + {"contents": []}, + "https://aiplatform.googleapis.com/v1/projects/p/locations/global" + "/publishers/google/models/gemini-3.8-flash:countTokens", + None, + "gemini-3.8-flash", + "vertex_ai", + ), + ( + {}, + "https://us-central1-aiplatform.googleapis.com/v1/projects/p/locations/us-central1/cachedContents", + None, + "", + "vertex_ai", + ), + ({"contents": []}, _RELAYED_GEMINI_COUNT_TOKENS_TARGET, "gemini", "gemini-3.8-flash", "gemini"), + ({"contents": []}, _RELAYED_VERTEX_COUNT_TOKENS_TARGET, None, "", None), + ({"contents": []}, _TRANSLATE_TARGET, None, "", None), + ({"contents": []}, _NON_GOOGLE_VERTEX_SHAPED_TARGET, None, "", None), + ({"contents": []}, "https://api.example.com/v1/models/gpt-x:countTokens", None, "", None), + ({"contents": []}, "https://api.example.com/v1/models/gpt-x:predict", None, "", None), + ({"contents": []}, "https://api.example.com/v1/models/gpt-x:thing", None, "", None), + (None, None, None, "", None), + ], +) +def test_build_passthrough_failure_request_payload_model_and_provider_precedence( + parsed_body: dict | None, + url: str | None, + custom_llm_provider: str | None, + expected_model: str, + expected_provider: str | None, +): + request_payload: Final = _build_passthrough_failure_request_payload( + parsed_body=parsed_body, + kwargs=None, + logging_obj=None, + custom_llm_provider=custom_llm_provider, + url=httpx.URL(url) if url else None, + builtin_provider_route=True, + ) + assert request_payload["model"] == expected_model + assert request_payload.get("custom_llm_provider") == expected_provider + + +@pytest.mark.parametrize( + "url_route, expected_model", + [ + (_VERTEX_FLASH_TARGET, "gemini-3.8-flash"), + (_VERTEX_FLASH_TARGET.replace(":generateContent", ":streamGenerateContent?alt=sse"), "gemini-3.8-flash"), + (_VERTEX_FLASH_TARGET.replace(":generateContent", ":predict"), "gemini-3.8-flash"), + (_VERTEX_FLASH_TARGET.replace(":generateContent", ":embedContent"), "gemini-3.8-flash"), + (_VERTEX_CLAUDE_RAW_PREDICT_TARGET, "vertex_ai/claude-opus-5"), + (_VERTEX_CLAUDE_RAW_PREDICT_TARGET.replace(":rawPredict", ":streamRawPredict"), "vertex_ai/claude-opus-5"), + ( + _VERTEX_CLAUDE_RAW_PREDICT_TARGET.replace(":rawPredict", ":streamRawPredict?alt=sse"), + "claude-opus-5", + ), + (_VERTEX_CLAUDE_RAW_PREDICT_TARGET.replace(":rawPredict", ":rawPredict?alt=sse"), "claude-opus-5"), + (f"{_VERTEX_FLASH_TARGET}?tag=rawPredict", "gemini-3.8-flash"), + (_VERTEX_SEARCH_TARGET, "vertex_ai/search_api"), + ("https://us-east5-aiplatform.googleapis.com/v1/projects/p/locations/us-east5/batchPredictionJobs", None), + ( + "https://us-east5-aiplatform.googleapis.com/v1/projects/p/locations/us-east5/batchPredictionJobs" + "?filter=:search", + None, + ), + ], +) +def test_vertex_failure_model_name_matches_success_handler_naming(url_route: str, expected_model: str | None): + assert VertexPassthroughLoggingHandler.model_from_url_route(url_route) == expected_model + + @pytest.mark.parametrize("client_metadata_key", ["litellm_metadata", "metadata"]) def test_passthrough_logs_the_resolved_deployment_model_info_over_the_request_body(client_metadata_key: str): """A provider route that resolved a router deployment stashes its model_info on request.state. That @@ -8598,7 +9181,9 @@ def test_passthrough_logs_the_resolved_deployment_model_info_over_the_request_bo the call to (LIT-1761: passthrough successes carried model_id="").""" mock_request = MagicMock(spec=Request) mock_request.method = "POST" - mock_request.url = httpx.URL("http://0.0.0.0:4000/vertex_ai/v1/projects/p/locations/global/publishers/google/models/gemini-3.8-flash:generateContent") + mock_request.url = httpx.URL( + "http://0.0.0.0:4000/vertex_ai/v1/projects/p/locations/global/publishers/google/models/gemini-3.8-flash:generateContent" + ) mock_request.headers = Headers({}) mock_request.scope = {} mock_request.state = SimpleNamespace( @@ -9668,12 +10253,14 @@ async def _drain_logging_worker(): await GLOBAL_LOGGING_WORKER.stop() yield + @pytest.fixture() def _vcr_outcome_gate(request, vcr): install_live_call_probe(request, vcr) yield record_vcr_outcome(request, vcr) + @pytest.mark.usefixtures("_drain_logging_worker", "_vcr_outcome_gate") def test_update_pass_through_route_updates_registry(): """ @@ -9763,9 +10350,7 @@ def mock_request(): return self._dict.values() class MockRequest: - def __init__( - self, headers=None, method="POST", request_body: Optional[dict] = None - ): + def __init__(self, headers=None, method="POST", request_body: Optional[dict] = None): self.headers = headers or {} self.query_params = QueryParams() self.method = method @@ -9873,9 +10458,9 @@ def test_pass_through_routes_support_all_methods(): path = route.path methods = set(route.methods) allowed = PROTOCOL_CONSTRAINED_PASS_THROUGH_ROUTES.get(path, expected_methods) - assert ( - methods == allowed - ), f"Route {path} does not support all methods. Supported: {methods}, Expected: {allowed}" + assert methods == allowed, ( + f"Route {path} does not support all methods. Supported: {methods}, Expected: {allowed}" + ) check_router_methods(llm_router) @@ -9905,19 +10490,11 @@ def test_is_bedrock_agent_runtime_route(): # Test agent runtime endpoints (should return True) assert _is_bedrock_agent_runtime_route("/knowledgebases/kb-123/retrieve") is True - assert ( - _is_bedrock_agent_runtime_route("/agents/knowledgebases/kb-123/retrieve") - is True - ) + assert _is_bedrock_agent_runtime_route("/agents/knowledgebases/kb-123/retrieve") is True # Test regular bedrock runtime endpoints (should return False) - assert ( - _is_bedrock_agent_runtime_route("/guardrail/test-id/version/1/apply") is False - ) - assert ( - _is_bedrock_agent_runtime_route("/model/cohere.command-r-v1:0/converse") - is False - ) + assert _is_bedrock_agent_runtime_route("/guardrail/test-id/version/1/apply") is False + assert _is_bedrock_agent_runtime_route("/model/cohere.command-r-v1:0/converse") is False assert _is_bedrock_agent_runtime_route("/some/random/endpoint") is False @@ -9984,9 +10561,7 @@ def test_custom_pricing_used_in_cost_calculation(): # Verify custom cost is different from standard cost (unless prices happen to match) # This confirms custom pricing is actually being applied - assert ( - custom_cost != standard_cost - ), "Custom pricing should produce different cost than standard pricing" + assert custom_cost != standard_cost, "Custom pricing should produce different cost than standard pricing" # Test 3: Custom pricing with cache_read_input_token_cost and input_cost_per_token_batches # This specifically tests the parameters that were causing the original issue @@ -10562,6 +11137,7 @@ def test_init_kwargs_client_metadata_cannot_mark_a_pass_through_call_as_a_silent assert "is_silent_experiment" not in json.loads(row["metadata"]) assert result["litellm_params"]["metadata"]["client_key"] == "kept" + @pytest.mark.asyncio async def test_direct_passthrough_keeps_mapped_baggage_tags_outside_upstream_body(): from litellm.proxy.common_utils.http_parsing_utils import read_request_body From 8ac9b757370a27f51de80a8dcc84b26f743fe583 Mon Sep 17 00:00:00 2001 From: "devin-ai-integration[bot]" <158243242+devin-ai-integration[bot]@users.noreply.github.com> Date: Fri, 30 Oct 2026 09:43:01 -0700 Subject: [PATCH 11/18] fix(model_prices): rolling registry audit, add Gemini batch and Azure Astra flex tiers (#49114) * fix(model_prices): add EmpirioLabs model rows from the provider catalog Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(model_prices): align EmpirioLabs rows with the registry endpoint vocabulary and schema Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(model_prices): absorb Gemini long-prompt batch rates and Azure GPT-6 Astra long-context flex rates Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(model_prices): drop EmpirioLabs rows from the registry audit --------- Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> --- ...odel_prices_and_context_window_backup.json | 36 ++++++++++++++----- model_prices_and_context_window.json | 36 ++++++++++++++----- 2 files changed, 54 insertions(+), 18 deletions(-) diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json index 0799a298942..28a8fe83fa2 100644 --- a/litellm/model_prices_and_context_window_backup.json +++ b/litellm/model_prices_and_context_window_backup.json @@ -8715,7 +8715,7 @@ "search_context_size_low": 0.014, "search_context_size_medium": 0.014 }, - "source": "https://prices.azure.com/api/retail/prices?$filter=serviceName%20eq%20'Foundry%20Models'%20and%20armRegionName%20eq%20'eastus'%20and%20priceType%20eq%20'Consumption'", + "source": "https://azure.microsoft.com/en-us/pricing/details/cognitive-services/openai-service/", "supported_endpoints": [ "/v1/chat/completions", "/v1/responses" @@ -8742,7 +8742,11 @@ "supports_tool_choice": true, "supports_vision": true, "supports_web_search": true, - "supports_xhigh_reasoning_effort": true + "supports_xhigh_reasoning_effort": true, + "cache_creation_input_token_cost_above_272k_tokens_flex": 1.25e-05, + "cache_read_input_token_cost_above_272k_tokens_flex": 1e-06, + "input_cost_per_token_above_272k_tokens_flex": 1e-05, + "output_cost_per_token_above_272k_tokens_flex": 3.75e-05 }, "azure/gpt-6-astra-2026-09-03": { "cache_creation_input_token_cost": 1.25e-05, @@ -8763,7 +8767,7 @@ "search_context_size_low": 0.014, "search_context_size_medium": 0.014 }, - "source": "https://prices.azure.com/api/retail/prices?$filter=serviceName%20eq%20'Foundry%20Models'%20and%20armRegionName%20eq%20'eastus'%20and%20priceType%20eq%20'Consumption'", + "source": "https://azure.microsoft.com/en-us/pricing/details/cognitive-services/openai-service/", "supported_endpoints": [ "/v1/chat/completions", "/v1/responses" @@ -8790,7 +8794,11 @@ "supports_tool_choice": true, "supports_vision": true, "supports_web_search": true, - "supports_xhigh_reasoning_effort": true + "supports_xhigh_reasoning_effort": true, + "cache_creation_input_token_cost_above_272k_tokens_flex": 1.25e-05, + "cache_read_input_token_cost_above_272k_tokens_flex": 1e-06, + "input_cost_per_token_above_272k_tokens_flex": 1e-05, + "output_cost_per_token_above_272k_tokens_flex": 3.75e-05 }, "azure/gpt-6-luna": { "cache_creation_input_token_cost": 1.25e-07, @@ -30856,8 +30864,9 @@ "output_cost_per_token": 0, "output_vector_size": 3072, "rpm": 10000, - "source": "https://ai.google.dev/gemini-api/docs/embeddings#model-versions", - "tpm": 10000000 + "source": "https://ai.google.dev/gemini-api/docs/pricing", + "tpm": 10000000, + "input_cost_per_token_batches": 7.5e-08 }, "gemini/gemini-embedding-2": { "input_cost_per_audio_token": 6.5e-06, @@ -31713,7 +31722,10 @@ "input_cost_per_token_flex": 6.25e-07, "output_cost_per_token_batches": 5e-06, "output_cost_per_token_flex": 5e-06, - "supports_url_context": true + "supports_url_context": true, + "cache_read_input_token_cost_above_200k_tokens_batches": 2.5e-07, + "input_cost_per_token_above_200k_tokens_batches": 1.25e-06, + "output_cost_per_token_above_200k_tokens_batches": 7.5e-06 }, "gemini/gemini-2.5-computer-use-preview-10-2025": { "deprecation_date": "2026-07-28", @@ -32280,7 +32292,10 @@ "web_search_billing_unit": "per_query", "google_maps_grounding_cost_per_query": 0.014, "input_cost_per_token_flex": 1e-06, - "output_cost_per_token_flex": 6e-06 + "output_cost_per_token_flex": 6e-06, + "cache_read_input_token_cost_above_200k_tokens_batches": 4e-07, + "input_cost_per_token_above_200k_tokens_batches": 2e-06, + "output_cost_per_token_above_200k_tokens_batches": 9e-06 }, "gemini/gemini-3.1-pro-preview-customtools": { "prompt_cache_min_tokens": 4096, @@ -32343,7 +32358,10 @@ "web_search_billing_unit": "per_query", "google_maps_grounding_cost_per_query": 0.014, "input_cost_per_token_flex": 1e-06, - "output_cost_per_token_flex": 6e-06 + "output_cost_per_token_flex": 6e-06, + "cache_read_input_token_cost_above_200k_tokens_batches": 4e-07, + "input_cost_per_token_above_200k_tokens_batches": 2e-06, + "output_cost_per_token_above_200k_tokens_batches": 9e-06 }, "gemini-3-flash-preview": { "cache_read_input_audio_token_cost": 1e-07, diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json index 0799a298942..28a8fe83fa2 100644 --- a/model_prices_and_context_window.json +++ b/model_prices_and_context_window.json @@ -8715,7 +8715,7 @@ "search_context_size_low": 0.014, "search_context_size_medium": 0.014 }, - "source": "https://prices.azure.com/api/retail/prices?$filter=serviceName%20eq%20'Foundry%20Models'%20and%20armRegionName%20eq%20'eastus'%20and%20priceType%20eq%20'Consumption'", + "source": "https://azure.microsoft.com/en-us/pricing/details/cognitive-services/openai-service/", "supported_endpoints": [ "/v1/chat/completions", "/v1/responses" @@ -8742,7 +8742,11 @@ "supports_tool_choice": true, "supports_vision": true, "supports_web_search": true, - "supports_xhigh_reasoning_effort": true + "supports_xhigh_reasoning_effort": true, + "cache_creation_input_token_cost_above_272k_tokens_flex": 1.25e-05, + "cache_read_input_token_cost_above_272k_tokens_flex": 1e-06, + "input_cost_per_token_above_272k_tokens_flex": 1e-05, + "output_cost_per_token_above_272k_tokens_flex": 3.75e-05 }, "azure/gpt-6-astra-2026-09-03": { "cache_creation_input_token_cost": 1.25e-05, @@ -8763,7 +8767,7 @@ "search_context_size_low": 0.014, "search_context_size_medium": 0.014 }, - "source": "https://prices.azure.com/api/retail/prices?$filter=serviceName%20eq%20'Foundry%20Models'%20and%20armRegionName%20eq%20'eastus'%20and%20priceType%20eq%20'Consumption'", + "source": "https://azure.microsoft.com/en-us/pricing/details/cognitive-services/openai-service/", "supported_endpoints": [ "/v1/chat/completions", "/v1/responses" @@ -8790,7 +8794,11 @@ "supports_tool_choice": true, "supports_vision": true, "supports_web_search": true, - "supports_xhigh_reasoning_effort": true + "supports_xhigh_reasoning_effort": true, + "cache_creation_input_token_cost_above_272k_tokens_flex": 1.25e-05, + "cache_read_input_token_cost_above_272k_tokens_flex": 1e-06, + "input_cost_per_token_above_272k_tokens_flex": 1e-05, + "output_cost_per_token_above_272k_tokens_flex": 3.75e-05 }, "azure/gpt-6-luna": { "cache_creation_input_token_cost": 1.25e-07, @@ -30856,8 +30864,9 @@ "output_cost_per_token": 0, "output_vector_size": 3072, "rpm": 10000, - "source": "https://ai.google.dev/gemini-api/docs/embeddings#model-versions", - "tpm": 10000000 + "source": "https://ai.google.dev/gemini-api/docs/pricing", + "tpm": 10000000, + "input_cost_per_token_batches": 7.5e-08 }, "gemini/gemini-embedding-2": { "input_cost_per_audio_token": 6.5e-06, @@ -31713,7 +31722,10 @@ "input_cost_per_token_flex": 6.25e-07, "output_cost_per_token_batches": 5e-06, "output_cost_per_token_flex": 5e-06, - "supports_url_context": true + "supports_url_context": true, + "cache_read_input_token_cost_above_200k_tokens_batches": 2.5e-07, + "input_cost_per_token_above_200k_tokens_batches": 1.25e-06, + "output_cost_per_token_above_200k_tokens_batches": 7.5e-06 }, "gemini/gemini-2.5-computer-use-preview-10-2025": { "deprecation_date": "2026-07-28", @@ -32280,7 +32292,10 @@ "web_search_billing_unit": "per_query", "google_maps_grounding_cost_per_query": 0.014, "input_cost_per_token_flex": 1e-06, - "output_cost_per_token_flex": 6e-06 + "output_cost_per_token_flex": 6e-06, + "cache_read_input_token_cost_above_200k_tokens_batches": 4e-07, + "input_cost_per_token_above_200k_tokens_batches": 2e-06, + "output_cost_per_token_above_200k_tokens_batches": 9e-06 }, "gemini/gemini-3.1-pro-preview-customtools": { "prompt_cache_min_tokens": 4096, @@ -32343,7 +32358,10 @@ "web_search_billing_unit": "per_query", "google_maps_grounding_cost_per_query": 0.014, "input_cost_per_token_flex": 1e-06, - "output_cost_per_token_flex": 6e-06 + "output_cost_per_token_flex": 6e-06, + "cache_read_input_token_cost_above_200k_tokens_batches": 4e-07, + "input_cost_per_token_above_200k_tokens_batches": 2e-06, + "output_cost_per_token_above_200k_tokens_batches": 9e-06 }, "gemini-3-flash-preview": { "cache_read_input_audio_token_cost": 1e-07, From 9432bf6b8ee686d0fbd06863e5f75889d383c8a2 Mon Sep 17 00:00:00 2001 From: tavren27 <348969968+tavren27@users.noreply.github.com> Date: Fri, 30 Oct 2026 10:01:44 -0700 Subject: [PATCH 12/18] perf(streaming): stop raising caught AttributeErrors inside pydantic on every streamed chunk (#49166) * perf(streaming): stop raising AttributeErrors inside pydantic on every streamed chunk Every later plain chunk on openai and azure streams raised and caught 12 AttributeErrors inside pydantic. Assigning an undeclared extra goes through a class lookup that raises in the model metaclass and is never memoized, and the hidden_params getter went through the private attribute fallback set_extra_attribute writes the extra and fields_set directly when a cached per-class check says pydantic would do the same, and falls back to setattr otherwise. The hidden_params getter reads the private storage directly. Emitted chunk JSON is unchanged * fix(streaming): match pydantic's own extra-field bookkeeping on every supported version Take the direct extra write only when the installed pydantic itself adds an assigned extra to model_fields_set (2.13 and later) and the class defines no attribute of that name, so pydantic 2.11, 2.12 and 2.14 see the same model state as a plain setattr. A deleted _hidden_params raises AttributeError again, and the mutable test models carry frozen-ok reasons. * fix(streaming): keep the normal write for models that do not allow extra fields openai's construct gives every model an extras dict, so a forbid model built that way still needs setattr's ValueError instead of a silent extra write. * test(streaming): cover a ClassVar declared without a value in the extra-field parity test * fix(streaming): type the streamed choice's logprobs as the parsed value main now stores --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> --- litellm/litellm_core_utils/core_helpers.py | 7 +- litellm/litellm_core_utils/hidden_params.py | 35 ++- .../litellm_core_utils/streaming_handler.py | 17 +- litellm/types/utils.py | 32 ++- .../litellm_core_utils/test_hidden_params.py | 139 +++++++++++- .../test_streaming_handler.py | 211 +++++++++++++++--- 6 files changed, 388 insertions(+), 53 deletions(-) diff --git a/litellm/litellm_core_utils/core_helpers.py b/litellm/litellm_core_utils/core_helpers.py index 2f81154ab83..20e10f63f83 100644 --- a/litellm/litellm_core_utils/core_helpers.py +++ b/litellm/litellm_core_utils/core_helpers.py @@ -9,12 +9,13 @@ from types import MappingProxyType from typing import TYPE_CHECKING, Any, Final, Literal, Protocol, TypeVar, cast, runtime_checkable import httpx -from pydantic import TypeAdapter, ValidationError +from pydantic import BaseModel, TypeAdapter, ValidationError from typing_extensions import TypeIs # noqa: TID251 # narrows a duck-typed value to the Protocol its caller reads import litellm from litellm._logging import verbose_logger from litellm.constants import KUBERNETES_POD_ROUTING_KEY +from litellm.litellm_core_utils.hidden_params import set_extra_attribute from litellm.litellm_core_utils.llm_response_utils.get_headers import get_provider_request_id from litellm.types.llms.openai import AllMessageValues, OpenAIChatCompletionFinishReason @@ -710,9 +711,9 @@ def preserve_upstream_non_openai_attributes( _assign_attributes(model_response, original_chunk.model_dump(include=frozenset(candidate_keys)).items()) -def _assign_attributes(target: object, values: Iterable[tuple[str, object]]) -> None: +def _assign_attributes(target: BaseModel, values: Iterable[tuple[str, object]]) -> None: for key, value in values: - setattr(target, key, value) + set_extra_attribute(target, key, value) def safe_deep_copy(data): diff --git a/litellm/litellm_core_utils/hidden_params.py b/litellm/litellm_core_utils/hidden_params.py index a74337d2549..c04d7614b90 100644 --- a/litellm/litellm_core_utils/hidden_params.py +++ b/litellm/litellm_core_utils/hidden_params.py @@ -1,8 +1,9 @@ from collections.abc import Iterator, Mapping, MutableMapping +from functools import cache, lru_cache from types import MappingProxyType from typing import Final, Protocol, cast -from pydantic import BaseModel +from pydantic import BaseModel, ConfigDict, create_model from litellm.types.llms.base import HiddenParams @@ -60,6 +61,38 @@ def stored_attribute(obj: object, name: str) -> object: return extras.get(name) +@cache +def _assigned_extras_join_fields_set() -> bool: + probe: Final = create_model("_ExtraAssignmentProbe", __config__=ConfigDict(extra="allow"))() + setattr(probe, "assigned", None) + return "assigned" in probe.model_fields_set + + +@lru_cache(maxsize=1024) +def _writes_plain_extra(model_class: type[BaseModel], name: str) -> bool: + config: Final = model_class.model_config + return ( + model_class.__setattr__ is BaseModel.__setattr__ + and config.get("extra") == "allow" + and not config.get("frozen") + and not config.get("validate_assignment") + and not name.startswith("_") + and name not in model_class.__pydantic_fields__ + and name not in model_class.__class_vars__ + and not hasattr(model_class, name) + and _assigned_extras_join_fields_set() + ) + + +def set_extra_attribute(model: BaseModel, name: str, value: object) -> None: + extras: Final = model.__pydantic_extra__ + if extras is None or not _writes_plain_extra(type(model), name): + setattr(model, name, value) + return + extras[name] = value + model.__pydantic_fields_set__.add(name) + + def _get_hidden_params_storage(obj: object) -> object | None: if isinstance(obj, dict): return obj.get(_HIDDEN_PARAMS_ATTR) diff --git a/litellm/litellm_core_utils/streaming_handler.py b/litellm/litellm_core_utils/streaming_handler.py index 098a404ea14..b1b3e20f218 100644 --- a/litellm/litellm_core_utils/streaming_handler.py +++ b/litellm/litellm_core_utils/streaming_handler.py @@ -28,6 +28,7 @@ from litellm.litellm_core_utils.core_helpers import chunk_usage, provider_report from litellm.litellm_core_utils.hidden_params import ( HIDDEN_PARAMS_ATTR, get_hidden_params, + set_extra_attribute, set_hidden_param, stored_attribute, ) @@ -1035,7 +1036,7 @@ class CustomStreamWrapper: _initial_delta: Final = model_response.choices[0].delta.model_dump() _initial_delta.pop("role", None) - model_response.choices[0].delta = Delta(**_initial_delta) + set_extra_attribute(model_response.choices[0], "delta", Delta(**_initial_delta)) return model_response def _prebuilt_plain_delta(self, chunk: object) -> Delta | None: @@ -1061,13 +1062,13 @@ class CustomStreamWrapper: def _set_role_on_fresh_delta(self, model_response: ModelResponseStream) -> None: delta: Final = model_response.choices[0].delta if self.sent_first_chunk is False: - delta.role = "assistant" + set_extra_attribute(delta, "role", "assistant") self.sent_first_chunk = True return if not _holds_only_scalars(delta): self.strip_role_from_delta(model_response) return - delta.role = None + set_extra_attribute(delta, "role", None) if "provider_specific_fields" not in delta.model_fields_set: delta.provider_specific_fields = None @@ -1186,8 +1187,8 @@ class CustomStreamWrapper: and _holds_only_scalars(upstream_delta) ) if can_reuse_choice and upstream_choice is not None: - model_response.choices[0].finish_reason = None # rebind-ok: keep reused choice - model_response.choices[0].index = upstream_choice.index # rebind-ok: keep reused choice + set_extra_attribute(model_response.choices[0], "finish_reason", None) + set_extra_attribute(model_response.choices[0], "index", upstream_choice.index) _prepend_upstream_choice_extras( model_response.choices[0], upstream_choice, @@ -1212,11 +1213,7 @@ class CustomStreamWrapper: else: return model_response.system_fingerprint = upstream_chunk.system_fingerprint - setattr( - model_response, - "citations", - stored_attribute(upstream_chunk, "citations"), - ) + set_extra_attribute(model_response, "citations", stored_attribute(upstream_chunk, "citations")) preserve_upstream_non_openai_attributes( model_response=model_response, original_chunk=original_chunk, diff --git a/litellm/types/utils.py b/litellm/types/utils.py index 55aaf606175..f8115571dfd 100644 --- a/litellm/types/utils.py +++ b/litellm/types/utils.py @@ -75,6 +75,7 @@ from litellm.types.workload_identity import ( ) from ..litellm_core_utils.core_helpers import map_finish_reason, process_response_headers +from ..litellm_core_utils.hidden_params import set_extra_attribute from . import litellm_params as _litellm_params from .agents import LiteLLMSendMessageResponse from .guardrails import GuardrailEventHooks @@ -2149,6 +2150,13 @@ class UnreportedUsage(Usage): class StreamingChoices(OpenAIObject): + if TYPE_CHECKING: + finish_reason: str | None + index: int + delta: Delta + enhancements: object + logprobs: ChoiceLogprobs | JsonValue + def __init__( self, finish_reason=None, @@ -2162,25 +2170,22 @@ class StreamingChoices(OpenAIObject): # https://github.com/BerriAI/litellm/issues/8455 params.pop("message", None) super().__init__(**params) - if finish_reason: - self.finish_reason = map_finish_reason(finish_reason) - else: - self.finish_reason = None - self.index = index + set_extra_attribute(self, "finish_reason", map_finish_reason(finish_reason) if finish_reason else None) + set_extra_attribute(self, "index", index) if delta is not None: if isinstance(delta, Delta): - self.delta = delta + set_extra_attribute(self, "delta", delta) elif isinstance(delta, dict): - self.delta = Delta(**delta) + set_extra_attribute(self, "delta", Delta(**delta)) else: - self.delta = Delta() + set_extra_attribute(self, "delta", Delta()) if enhancements is not None: - self.enhancements = enhancements + set_extra_attribute(self, "enhancements", enhancements) if isinstance(logprobs, (dict, BaseModel)): - self.logprobs = CHOICE_LOGPROBS_OR_JSON.validate_python(logprobs) + set_extra_attribute(self, "logprobs", CHOICE_LOGPROBS_OR_JSON.validate_python(logprobs)) else: - self.logprobs = logprobs + set_extra_attribute(self, "logprobs", logprobs) def __contains__(self, key) -> bool: # Define custom behavior for the 'in' operator @@ -2234,7 +2239,10 @@ class ModelResponseBase(OpenAIObject): @property def hidden_params(self) -> dict[str, object]: # mutable-ok: API requires mutation - return self._hidden_params + private: Final = self.__pydantic_private__ + if private is None or "_hidden_params" not in private: + return self._hidden_params + return private["_hidden_params"] @hidden_params.setter def hidden_params(self, hidden_params: dict[str, object]) -> None: # mutable-ok: API requires mutation diff --git a/tests/unit/litellm_core_utils/test_hidden_params.py b/tests/unit/litellm_core_utils/test_hidden_params.py index c0485a07696..b32692c67ba 100644 --- a/tests/unit/litellm_core_utils/test_hidden_params.py +++ b/tests/unit/litellm_core_utils/test_hidden_params.py @@ -1,7 +1,11 @@ -from typing import Final +import pickle +from collections.abc import Callable +from typing import ClassVar, Final import pytest +from openai import BaseModel as OpenAIBaseModel from openai.types.chat import ChatCompletionChunk +from pydantic import BaseModel, ConfigDict, PrivateAttr from litellm.litellm_core_utils.hidden_params import ( HIDDEN_PARAMS_ATTR, @@ -9,13 +13,14 @@ from litellm.litellm_core_utils.hidden_params import ( get_hidden_params_storage, get_or_create_hidden_params, served_from_cache, + set_extra_attribute, set_hidden_param, set_hidden_params, ) from litellm.llms.openai.completion.transformation import OpenAITextCompletionConfig from litellm.types.decisions import DecisionsResponse from litellm.types.llms.base import HiddenParams -from litellm.types.utils import ModelResponse, TextChoices, TextCompletionResponse, Usage +from litellm.types.utils import ModelResponse, ModelResponseStream, TextChoices, TextCompletionResponse, Usage def test_get_and_set_hidden_params_on_plain_object() -> None: @@ -309,3 +314,133 @@ def test_served_from_cache_is_true_only_for_a_response_the_cache_stamped( assert served_from_cache(model_response) is expected assert served_from_cache({HIDDEN_PARAMS_ATTR: dict(hidden_params)}) is expected assert served_from_cache(object()) is False + + +class _ExtraModel(BaseModel): # frozen-ok: the assignment tests need a model that accepts attribute writes + model_config = ConfigDict(extra="allow") + + declared: str | None = None + marker: ClassVar[str | None] = None + unset_marker: ClassVar[int] + _secret: str = PrivateAttr(default="hidden") + + @property + def doubled(self) -> str | None: + return self.declared + + @doubled.setter + def doubled(self, value: str) -> None: + self.declared = value * 2 + + +class _FrozenExtraModel(BaseModel): + model_config = ConfigDict(extra="allow", frozen=True) + + +class _ValidatedExtraModel(BaseModel): # frozen-ok: the assignment tests need a model that accepts attribute writes + model_config = ConfigDict(extra="allow", validate_assignment=True) + + __pydantic_extra__: dict[str, int] + + +class _IgnoredExtraModel(BaseModel): # frozen-ok: the assignment tests need a model that accepts attribute writes + model_config = ConfigDict(extra="ignore") + + +class _MirroringModel(BaseModel): # frozen-ok: the assignment tests need a model that accepts attribute writes + model_config = ConfigDict(extra="allow") + + def __setattr__(self, name: str, value: object) -> None: + super().__setattr__(name, value) + super().__setattr__(f"{name}_copy", value) + + +class _NoneClassAttribute: + inherited = None + + +class _InheritingExtraModel(_NoneClassAttribute, BaseModel): # frozen-ok: assignment tests need attribute writes + model_config = ConfigDict(extra="allow") + + +class _ForbiddingOpenAIModel(OpenAIBaseModel): # frozen-ok: assignment tests need attribute writes + model_config = ConfigDict(extra="forbid") + + +def _assignment_outcome(model: BaseModel, assign: Callable[[], None], name: str) -> tuple[object, ...]: + try: + assign() + except (AttributeError, ValueError) as error: + return type(error), str(error) + return model.model_dump(), getattr(model, name), model.model_fields_set, model.model_dump_json() + + +@pytest.mark.parametrize( + "build, name, value", + [ + pytest.param(lambda: _ExtraModel(existing=1, later=3), "fresh", 1, id="new-extra"), + pytest.param(lambda: _ExtraModel(existing=1, later=3), "existing", 2, id="existing-extra-keeps-order"), + pytest.param(lambda: _ExtraModel(existing=1, later=3), "declared", "x", id="declared-field"), + pytest.param(lambda: _ExtraModel(existing=1, later=3), "doubled", "ab", id="property"), + pytest.param(lambda: _ExtraModel(existing=1, later=3), "_free", 1, id="undeclared-underscore-name"), + pytest.param(lambda: _ExtraModel(existing=1, later=3), "_secret", "x", id="private-attribute"), + pytest.param(lambda: _ExtraModel(existing=1, later=3), "marker", 1, id="class-var"), + pytest.param(lambda: _ExtraModel(existing=1, later=3), "unset_marker", 1, id="class-var-without-value"), + pytest.param(lambda: _ExtraModel(existing=1, later=3), "model_config", 1, id="class-attribute"), + pytest.param(_FrozenExtraModel, "fresh", 1, id="frozen"), + pytest.param(_ValidatedExtraModel, "fresh", "5", id="validate-assignment"), + pytest.param(_IgnoredExtraModel, "fresh", 1, id="extra-ignore"), + pytest.param(_MirroringModel, "fresh", 1, id="setattr-override"), + pytest.param(_InheritingExtraModel, "inherited", 1, id="class-attribute-set-to-none"), + pytest.param( + lambda: _ForbiddingOpenAIModel.construct(existing=1), "fresh", 1, id="extra-forbid-built-by-construct" + ), + ], +) +def test_set_extra_attribute_matches_setattr(build: Callable[[], BaseModel], name: str, value: object) -> None: + through_setattr: Final = build() + through_helper: Final = build() + + expected: Final = _assignment_outcome(through_setattr, lambda: setattr(through_setattr, name, value), name) + actual: Final = _assignment_outcome(through_helper, lambda: set_extra_attribute(through_helper, name, value), name) + + assert actual == expected + + +@pytest.mark.parametrize("response_type", [ModelResponse, ModelResponseStream]) +def test_hidden_params_raises_attribute_error_once_deleted( + response_type: type[ModelResponse | ModelResponseStream], +) -> None: + response: Final = response_type() + del response._hidden_params + + assert getattr(response, "hidden_params", "fallback") == "fallback" + + +def _round_trip(response: ModelResponse | ModelResponseStream) -> ModelResponse | ModelResponseStream: + return pickle.loads(pickle.dumps(response)) + + +@pytest.mark.parametrize("response_type", [ModelResponse, ModelResponseStream]) +@pytest.mark.parametrize( + "rebuild", + [ + pytest.param(lambda response: response, id="init"), + pytest.param(lambda response: response.model_copy(), id="model-copy"), + pytest.param(lambda response: response.model_copy(deep=True), id="deep-copy"), + pytest.param(_round_trip, id="pickle"), + ], +) +def test_hidden_params_is_the_stored_dict_on_every_construction_path( + response_type: type[ModelResponse | ModelResponseStream], + rebuild: Callable[[ModelResponse | ModelResponseStream], ModelResponse | ModelResponseStream], +) -> None: + original: Final = response_type() + original.hidden_params["model_id"] = "deployment-1" + response: Final = rebuild(original) + + response.hidden_params["api_base"] = "https://example.com" + + assert response.hidden_params is response._hidden_params + assert get_hidden_params(response) == response._hidden_params + assert response._hidden_params["api_base"] == "https://example.com" diff --git a/tests/unit/litellm_core_utils/test_streaming_handler.py b/tests/unit/litellm_core_utils/test_streaming_handler.py index b9c8feb223d..148b7193ef6 100644 --- a/tests/unit/litellm_core_utils/test_streaming_handler.py +++ b/tests/unit/litellm_core_utils/test_streaming_handler.py @@ -11,7 +11,7 @@ import time from collections.abc import Mapping, Sequence from dataclasses import dataclass from pathlib import Path -from types import CodeType, FrameType, MappingProxyType +from types import CodeType, FrameType, MappingProxyType, TracebackType from typing import AsyncIterator, Callable, Final, Iterator, Literal, NoReturn, Optional, cast from unittest.mock import AsyncMock, MagicMock, Mock, patch @@ -6292,53 +6292,76 @@ def test_openai_chunk_creator_skips_full_chunk_dump_and_delta_rebuild( assert finish_output.choices[0].delta.content is None -def _openai_chunk(chunk_id: str, delta: Mapping[str, object]) -> ChatCompletionChunk: +def _openai_chunk( + chunk_id: str, + delta: Mapping[str, object], + choice_extras: Mapping[str, object] = MappingProxyType({}), + finish_reason: str | None = None, + chunk_extras: Mapping[str, object] = MappingProxyType({}), +) -> ChatCompletionChunk: return ChatCompletionChunk( id=chunk_id, created=1, model="gpt-4o", object="chat.completion.chunk", - choices=[{"index": 0, "delta": delta, "finish_reason": None}], + choices=[{"index": 0, "delta": delta, "finish_reason": finish_reason, **choice_extras}], + **chunk_extras, ) -def _missed_attributes(wrapper: CustomStreamWrapper, chunks: Sequence[ChatCompletionChunk]) -> tuple[str, ...]: - attribute_lookup: Final = BaseModel.__getattr__.__code__ - misses: Final[list[str]] = [] +def _raised_attribute_errors(wrapper: CustomStreamWrapper, chunks: Sequence[ChatCompletionChunk]) -> tuple[str, ...]: + raised: Final[list[str]] = [] - def record_miss(frame: FrameType, event: str, arg: object) -> Callable[[FrameType, str, object], object]: - if event == "exception" and isinstance(arg, tuple) and isinstance(arg[1], AttributeError): - misses.append(str(frame.f_locals["item"])) - return record_miss - - def trace_lookups(frame: FrameType, event: str, _arg: object) -> Callable[[FrameType, str, object], object] | None: - return record_miss if event == "call" and frame.f_code is attribute_lookup else None + def record_raise(frame: FrameType, event: str, arg: object) -> Callable[[FrameType, str, object], object]: + if event == "exception" and isinstance(arg, tuple): + _, error, traceback = arg + if isinstance(error, AttributeError) and isinstance(traceback, TracebackType) and traceback.tb_next is None: + raised.append(f"{frame.f_code.co_qualname}: {error}") + return record_raise previous_tracer: Final = sys.gettrace() - sys.settrace(trace_lookups) + sys.settrace(record_raise) try: tuple(wrapper.chunk_creator(chunk=chunk) for chunk in chunks) finally: sys.settrace(previous_tracer) - return tuple(misses) + return tuple(raised) -def test_openai_chunk_creator_builds_one_delta_and_misses_no_attribute_per_later_chunk( - logging_obj: Logging, +_AZURE_CHOICE_EXTRAS: Final = MappingProxyType( + {"content_filter_results": {"hate": {"filtered": False, "severity": "safe"}}} +) +_OPENAI_AND_AZURE_CHOICE_EXTRAS: Final = ( + pytest.param("openai", MappingProxyType({}), id="openai"), + pytest.param("azure", _AZURE_CHOICE_EXTRAS, id="azure"), +) + + +@pytest.mark.parametrize(("custom_llm_provider", "choice_extras"), _OPENAI_AND_AZURE_CHOICE_EXTRAS) +def test_openai_chunk_creator_builds_one_delta_and_raises_no_attribute_error_per_later_chunk( + logging_obj: Logging, custom_llm_provider: str, choice_extras: Mapping[str, object] ) -> None: wrapper: Final = CustomStreamWrapper( completion_stream=iter(()), model="gpt-4o", - custom_llm_provider="openai", + custom_llm_provider=custom_llm_provider, logging_obj=logging_obj, ) - wrapper.chunk_creator(chunk=_openai_chunk("chunk-1", {"role": "assistant", "content": ""})) - plain_chunks: Final = ( - _openai_chunk("chunk-2", {"content": "hello"}), - _openai_chunk("chunk-3", {"content": " world"}), + + def later_chunk(chunk_id: str, delta: Mapping[str, object]) -> ChatCompletionChunk: + return _openai_chunk( + chunk_id, delta, choice_extras, chunk_extras={"service_tier": "default", "obfuscation": chunk_id} + ) + + nested_delta: Final = {"content": "!", "reasoning_details": [{"type": "reasoning.text", "text": "thought"}]} + first_chunk_of_each_shape: Final = ( + later_chunk("chunk-1", {"role": "assistant", "content": ""}), + later_chunk("chunk-2", nested_delta), ) - nested_chunk: Final = _openai_chunk( - "chunk-4", {"content": "!", "reasoning_details": [{"type": "reasoning.text", "text": "thought"}]} + tuple(wrapper.chunk_creator(chunk=chunk) for chunk in first_chunk_of_each_shape) + plain_chunks: Final = ( + later_chunk("chunk-3", {"content": "hello"}), + later_chunk("chunk-4", {"content": " world"}), ) delta_init: Final = Delta.__init__.__code__ streaming_choices_init: Final = StreamingChoices.__init__.__code__ @@ -6348,7 +6371,145 @@ def test_openai_chunk_creator_builds_one_delta_and_misses_no_attribute_per_later assert sum(call.code is delta_init for call in calls) == len(plain_chunks) assert sum(call.code is streaming_choices_init for call in calls) == len(plain_chunks) assert [output.choices[0].delta.content if output else None for output in outputs] == ["hello", " world"] - assert _missed_attributes(wrapper, (*plain_chunks, nested_chunk)) == () + assert _raised_attribute_errors(wrapper, (*plain_chunks, later_chunk("chunk-5", nested_delta))) == () + + +_OPENAI_STREAM_JSON: Final = ( + ( + '{"id":"chatcmpl-1","created":1,"model":"gpt-4o","object":"chat.completion.chunk",' + '"system_fingerprint":null,"choices":[{"finish_reason":null,"index":0,' + '"delta":{"provider_specific_fields":null,"refusal":null,"content":"","role":"assistant",' + '"function_call":null,"tool_calls":null,"audio":null},"logprobs":null}],' + '"provider_specific_fields":null,"citations":null,"service_tier":"default","obfuscation":"o0"}' + ), + ( + '{"id":"chatcmpl-1","created":1,"model":"gpt-4o","object":"chat.completion.chunk",' + '"system_fingerprint":null,"choices":[{"finish_reason":null,"index":0,' + '"delta":{"provider_specific_fields":null,"refusal":null,"content":"one","role":null,' + '"function_call":null,"tool_calls":null,"audio":null},"logprobs":null}],' + '"provider_specific_fields":null,"citations":null,"service_tier":"default","obfuscation":"o1"}' + ), + ( + '{"id":"chatcmpl-1","created":1,"model":"gpt-4o","object":"chat.completion.chunk",' + '"system_fingerprint":null,"choices":[{"finish_reason":null,"index":0,' + '"delta":{"provider_specific_fields":null,"refusal":null,"content":" two","role":null,' + '"function_call":null,"tool_calls":null,"audio":null},"logprobs":null}],' + '"provider_specific_fields":null,"citations":null,"service_tier":"default","obfuscation":"o2"}' + ), + ( + '{"id":"chatcmpl-1","created":1,"model":"gpt-4o","object":"chat.completion.chunk",' + '"system_fingerprint":null,"choices":[{"finish_reason":"stop","index":0,' + '"delta":{"provider_specific_fields":null,"content":null,"role":null,"function_call":null,' + '"tool_calls":null,"audio":null},"logprobs":null}],"provider_specific_fields":null,' + '"service_tier":"default","obfuscation":"o3"}' + ), + ( + '{"id":"chatcmpl-1","created":1,"model":"gpt-4o","object":"chat.completion.chunk",' + '"system_fingerprint":null,"choices":[],"provider_specific_fields":null,' + '"usage":{"completion_tokens":3,"prompt_tokens":9,"total_tokens":12,"completion_tokens_details":null,' + '"prompt_tokens_details":null}}' + ), +) +_AZURE_STREAM_JSON: Final = ( + ( + '{"id":"chatcmpl-1","created":1,"model":"gpt-4o","object":"chat.completion.chunk",' + '"system_fingerprint":null,"choices":[{"content_filter_results":{"hate":{"filtered":false,' + '"severity":"safe"}},"finish_reason":null,"index":0,"delta":{"provider_specific_fields":null,' + '"refusal":null,"content":"","role":"assistant","function_call":null,"tool_calls":null,"audio":null},' + '"logprobs":null}],"provider_specific_fields":null,"citations":null,"service_tier":"default",' + '"obfuscation":"o0"}' + ), + ( + '{"id":"chatcmpl-1","created":1,"model":"gpt-4o","object":"chat.completion.chunk",' + '"system_fingerprint":null,"choices":[{"content_filter_results":{"hate":{"filtered":false,' + '"severity":"safe"}},"finish_reason":null,"index":0,"delta":{"provider_specific_fields":null,' + '"refusal":null,"content":"one","role":null,"function_call":null,"tool_calls":null,"audio":null},' + '"logprobs":null}],"provider_specific_fields":null,"citations":null,"service_tier":"default",' + '"obfuscation":"o1"}' + ), + ( + '{"id":"chatcmpl-1","created":1,"model":"gpt-4o","object":"chat.completion.chunk",' + '"system_fingerprint":null,"choices":[{"content_filter_results":{"hate":{"filtered":false,' + '"severity":"safe"}},"finish_reason":null,"index":0,"delta":{"provider_specific_fields":null,' + '"refusal":null,"content":" two","role":null,"function_call":null,"tool_calls":null,"audio":null},' + '"logprobs":null}],"provider_specific_fields":null,"citations":null,"service_tier":"default",' + '"obfuscation":"o2"}' + ), + ( + '{"id":"chatcmpl-1","created":1,"model":"gpt-4o","object":"chat.completion.chunk",' + '"system_fingerprint":null,"choices":[{"finish_reason":"stop","index":0,' + '"delta":{"provider_specific_fields":null,"content":null,"role":null,"function_call":null,' + '"tool_calls":null,"audio":null},"logprobs":null}],"provider_specific_fields":null,' + '"service_tier":"default","obfuscation":"o3"}' + ), + ( + '{"id":"chatcmpl-1","created":1,"model":"gpt-4o","object":"chat.completion.chunk",' + '"system_fingerprint":null,"choices":[],"provider_specific_fields":null,' + '"usage":{"completion_tokens":3,"prompt_tokens":9,"total_tokens":12,"completion_tokens_details":null,' + '"prompt_tokens_details":null}}' + ), +) + + +@pytest.mark.parametrize( + ("custom_llm_provider", "choice_extras", "expected_json"), + [ + pytest.param("openai", MappingProxyType({}), _OPENAI_STREAM_JSON, id="openai"), + pytest.param("azure", _AZURE_CHOICE_EXTRAS, _AZURE_STREAM_JSON, id="azure"), + ], +) +def test_openai_chunk_creator_emits_each_chunk_as_the_same_json( + logging_obj: Logging, + custom_llm_provider: str, + choice_extras: Mapping[str, object], + expected_json: tuple[str | None, ...], +) -> None: + wrapper: Final = CustomStreamWrapper( + completion_stream=iter(()), + model="gpt-4o", + custom_llm_provider=custom_llm_provider, + logging_obj=logging_obj, + stream_options={"include_usage": True}, + ) + chunks: Final = ( + _openai_chunk( + "chatcmpl-1", + {"role": "assistant", "content": ""}, + choice_extras, + chunk_extras={"service_tier": "default", "obfuscation": "o0"}, + ), + _openai_chunk( + "chatcmpl-1", + {"content": "one"}, + choice_extras, + chunk_extras={"service_tier": "default", "obfuscation": "o1"}, + ), + _openai_chunk( + "chatcmpl-1", + {"content": " two"}, + choice_extras, + chunk_extras={"service_tier": "default", "obfuscation": "o2"}, + ), + _openai_chunk( + "chatcmpl-1", {}, choice_extras, "stop", chunk_extras={"service_tier": "default", "obfuscation": "o3"} + ), + ChatCompletionChunk( + id="chatcmpl-1", + created=1, + model="gpt-4o", + object="chat.completion.chunk", + choices=[], + service_tier="default", + obfuscation="o4", + usage={"prompt_tokens": 9, "completion_tokens": 3, "total_tokens": 12}, + ), + ) + + outputs: Final = tuple(wrapper.chunk_creator(chunk=chunk) for chunk in chunks) + + assert tuple( + output.model_copy(update={"created": 1}).model_dump_json() if output else None for output in outputs + ) == expected_json def test_chunk_creator_skips_full_dump_of_model_response_stream_chunks( From ab3f6caddcd096b820359a7ae2829e9440b77c4f Mon Sep 17 00:00:00 2001 From: "devin-ai-integration[bot]" <158243242+devin-ai-integration[bot]@users.noreply.github.com> Date: Fri, 30 Oct 2026 10:18:29 -0700 Subject: [PATCH 13/18] fix(gemini): flag computer use support on the four models the Gemini computer use docs list (#49252) Co-authored-by: devin-ai-integration[bot] <158243242+devin-ai-integration[bot]@users.noreply.github.com> --- litellm/model_prices_and_context_window_backup.json | 10 +++++++--- model_prices_and_context_window.json | 10 +++++++--- 2 files changed, 14 insertions(+), 6 deletions(-) diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json index 28a8fe83fa2..e1131117fcf 100644 --- a/litellm/model_prices_and_context_window_backup.json +++ b/litellm/model_prices_and_context_window_backup.json @@ -31883,7 +31883,8 @@ }, "search_free_requests": {"count": 5000, "period": "month", "shared_group": "gemini-3"}, "web_search_billing_unit": "per_query", - "google_maps_grounding_cost_per_query": 0.014 + "google_maps_grounding_cost_per_query": 0.014, + "supports_computer_use": true }, "gemini/gemini-3-flash-preview": { "cache_read_input_audio_token_cost": 1e-07, @@ -31949,7 +31950,8 @@ "input_cost_per_token_flex": 2.5e-07, "output_cost_per_token_batches": 1.5e-06, "output_cost_per_token_flex": 1.5e-06, - "supports_audio_input": true + "supports_audio_input": true, + "supports_computer_use": true }, "gemini/gemini-3.5-flash": { "prompt_cache_min_tokens": 4096, @@ -32196,7 +32198,8 @@ }, "search_free_requests": {"count": 5000, "period": "month", "shared_group": "gemini-3"}, "web_search_billing_unit": "per_query", - "google_maps_grounding_cost_per_query": 0.014 + "google_maps_grounding_cost_per_query": 0.014, + "supports_computer_use": true }, "gemini/gemini-omni-flash-preview": { "input_cost_per_audio_token": 1.5e-06, @@ -76558,6 +76561,7 @@ ], "supports_audio_input": true, "supports_audio_output": false, + "supports_computer_use": true, "supports_function_calling": true, "supports_minimal_reasoning_effort": false, "supports_native_streaming": true, diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json index 28a8fe83fa2..e1131117fcf 100644 --- a/model_prices_and_context_window.json +++ b/model_prices_and_context_window.json @@ -31883,7 +31883,8 @@ }, "search_free_requests": {"count": 5000, "period": "month", "shared_group": "gemini-3"}, "web_search_billing_unit": "per_query", - "google_maps_grounding_cost_per_query": 0.014 + "google_maps_grounding_cost_per_query": 0.014, + "supports_computer_use": true }, "gemini/gemini-3-flash-preview": { "cache_read_input_audio_token_cost": 1e-07, @@ -31949,7 +31950,8 @@ "input_cost_per_token_flex": 2.5e-07, "output_cost_per_token_batches": 1.5e-06, "output_cost_per_token_flex": 1.5e-06, - "supports_audio_input": true + "supports_audio_input": true, + "supports_computer_use": true }, "gemini/gemini-3.5-flash": { "prompt_cache_min_tokens": 4096, @@ -32196,7 +32198,8 @@ }, "search_free_requests": {"count": 5000, "period": "month", "shared_group": "gemini-3"}, "web_search_billing_unit": "per_query", - "google_maps_grounding_cost_per_query": 0.014 + "google_maps_grounding_cost_per_query": 0.014, + "supports_computer_use": true }, "gemini/gemini-omni-flash-preview": { "input_cost_per_audio_token": 1.5e-06, @@ -76558,6 +76561,7 @@ ], "supports_audio_input": true, "supports_audio_output": false, + "supports_computer_use": true, "supports_function_calling": true, "supports_minimal_reasoning_effort": false, "supports_native_streaming": true, From 372c096388f3102419e0177f501f42126c7170c7 Mon Sep 17 00:00:00 2001 From: Etienne Chabert Date: Fri, 30 Oct 2026 18:20:00 +0100 Subject: [PATCH 14/18] fix(proxy): evict a deleted key from every worker's auth cache (#47609) * fix(proxy): evict a deleted key from every worker's auth cache /key/delete dropped the key object only from the in-memory cache of the worker that served the call. Auth serves key objects cache-first, so every other worker, and the Redis copy when enable_redis_auth_cache is on, kept authenticating the deleted key until user_api_key_cache_ttl expired. /key/update already evicts and broadcasts through evict_and_broadcast, delete now does the same for the tokens it deletes * fix(proxy): evict only the deleted rows and broadcast them in one batch Scope the cross-worker eviction on /key/delete to the rows the caller was allowed to delete and send it through delete_cache_key_objects, so a caller-supplied name can no longer drop another team's cached entries on every worker * refactor(proxy): drop the comment above the delete eviction call * fix(proxy): run the usage cache eviction beside the auth eviction so a hung Redis costs one wait * test(proxy): wait on the broadcast instead of a sleep and type the peer eviction tests * test(proxy): wait for the peer to apply the broadcast instead of polling its cache * test(proxy): cover hashed keys and aliases when deleting a key warmed on a peer * fix(key_management): evict every requested key on delete, including keys whose row is already gone A batch delete that names a live key and a key whose row was removed outside the API evicted only the live key, so the gone key kept authenticating from every worker's cache. Every requested key hash is evicted now, while raw non-key cache keys are still left alone * test(key_management): split the gone-row eviction assertion into one comprehension per worker --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> --- litellm/proxy/auth/auth_checks.py | 37 +-- .../key_management_endpoints.py | 15 +- .../management/test_key_delete_revocation.py | 15 +- tests/unit/proxy/auth/test_auth_checks.py | 50 +++- .../test_key_management_endpoints.py | 231 +++++++++++++++++- 5 files changed, 317 insertions(+), 31 deletions(-) diff --git a/litellm/proxy/auth/auth_checks.py b/litellm/proxy/auth/auth_checks.py index f80296f4bd7..79a46ef355e 100644 --- a/litellm/proxy/auth/auth_checks.py +++ b/litellm/proxy/auth/auth_checks.py @@ -3457,28 +3457,35 @@ async def delete_cache_key_objects( proxy_logging_obj: ProxyLogging | None, ) -> None: """ - Evict a batch of key objects, for callers that delete keys in bulk rather than through - `/key/delete`. Auth resolves a cached key object without re-reading its team, so a key left - cached after its row is gone keeps buying access until its TTL expires. - - Evicting locally only reaches this worker; `_delete_cache_key_object` itself broadcasts each - token, so a deleted key left in a peer worker's in-memory cache still authenticates there until - its TTL expires. + Evict a batch of deleted key objects here and on every other worker. Auth resolves a cached key + object without re-reading its team, so a key left cached after its row is gone keeps buying + access until its TTL expires. Best-effort per key: the rows are already deleted by the time this runs, so an unreachable cache backend must not abort the caller partway through its own cascade. - Every token is dropped from this worker's memory first, then from the auth cache and the - internal usage cache in one batch each, so a deleted key stops authenticating here the instant - its rows are gone instead of refilling from Redis on the next miss. Every other worker then - gets the eviction in one pipelined broadcast: a publish per key would pass the publisher's - pending cap on a bulk delete, and the dropped keys would keep authenticating there until their - TTL. + Every token is dropped from this worker's memory first, then from the auth cache's Redis layer, + so a deleted key stops authenticating here the instant its rows are gone instead of refilling + from Redis on the next miss. Every other worker then gets the eviction in one pipelined + broadcast: a publish per key would pass the publisher's pending cap on a bulk delete, and the + dropped keys would keep authenticating there until their TTL. The internal usage cache is not + an auth refill source, so its batch delete runs beside that chain and a hung Redis costs one + bounded wait there instead of another in line. """ evict_local(cache_keys=hashed_tokens, user_api_key_cache=user_api_key_cache) + usage_eviction: Final = ( + () + if proxy_logging_obj is None + else (evict_shared(cache_keys=hashed_tokens, cache=proxy_logging_obj.internal_usage_cache.dual_cache),) + ) + _ = await asyncio.gather( + _evict_auth_cache_then_broadcast(hashed_tokens=hashed_tokens, user_api_key_cache=user_api_key_cache), + *usage_eviction, + ) + + +async def _evict_auth_cache_then_broadcast(hashed_tokens: Sequence[str], user_api_key_cache: UserApiKeyCache) -> None: await evict_shared(cache_keys=hashed_tokens, cache=user_api_key_cache) - if proxy_logging_obj is not None: - await evict_shared(cache_keys=hashed_tokens, cache=proxy_logging_obj.internal_usage_cache.dual_cache) await broadcast_evictions(cache_keys=hashed_tokens) diff --git a/litellm/proxy/management_endpoints/key_management_endpoints.py b/litellm/proxy/management_endpoints/key_management_endpoints.py index a6814b593ca..3021b21de51 100644 --- a/litellm/proxy/management_endpoints/key_management_endpoints.py +++ b/litellm/proxy/management_endpoints/key_management_endpoints.py @@ -50,6 +50,7 @@ from litellm.proxy.auth.auth_checks import ( # noqa: F401 # legacy module expo _delete_cache_key_object, # pyright: ignore[reportPrivateUsage,reportUnusedImport] # backwards-compatible package export can_team_access_model, delete_cache_key_object, + delete_cache_key_objects, get_jwt_key_mapping_cache_keys_for_token, get_key_end_user_budget_id, get_org_object, @@ -78,7 +79,7 @@ from litellm.proxy.common_utils.config_sync_pubsub import ( from litellm.proxy.common_utils.openai_error_payload import openai_error_param, proxy_exception_for from litellm.proxy.common_utils.rbac_utils import check_org_admin_can_generate_keys from litellm.proxy.common_utils.timezone_utils import get_budget_reset_time -from litellm.proxy.common_utils.user_api_key_cache import AUTH_OBJECTS_TARGET, UserApiKeyCache +from litellm.proxy.common_utils.user_api_key_cache import AUTH_OBJECTS_TARGET, UserApiKeyCache, is_user_key_cache_key from litellm.proxy.hooks.key_management_event_hooks import KeyManagementEventHooks from litellm.proxy.hooks.model_max_budget_limiter import build_model_max_budget_usage from litellm.proxy.management.teams.authz import TEAM_ADMIN_ONLY, TEAM_OR_ORG_ADMIN, is_team_admin @@ -5483,7 +5484,7 @@ async def delete_verification_tokens( - List of keys being deleted, this contains information about the key_alias, token, and user_id being deleted, this is passed down to the KeyManagementEventHooks to delete the keys from the secret manager and handle audit logs """ - from litellm.proxy.proxy_server import prisma_client + from litellm.proxy.proxy_server import prisma_client, proxy_logging_obj failed_tokens: list = [] try: @@ -5567,11 +5568,11 @@ async def delete_verification_tokens( verbose_proxy_logger.debug(traceback.format_exc()) raise e - for key in tokens: - user_api_key_cache.delete_cache(key) - # remove hash token from cache - hashed_token = hash_token(cast(str, key)) - user_api_key_cache.delete_cache(hashed_token) + await delete_cache_key_objects( + hashed_tokens=tuple(token for token in hashed_tokens if is_user_key_cache_key(token)), + user_api_key_cache=user_api_key_cache, + proxy_logging_obj=proxy_logging_obj, + ) # After credential invalidation, so a failure here can never keep a deleted key alive. for deleted_key in authorized_keys: diff --git a/tests/integration/management/test_key_delete_revocation.py b/tests/integration/management/test_key_delete_revocation.py index 5b3527a52ae..bdc91dfe5f6 100644 --- a/tests/integration/management/test_key_delete_revocation.py +++ b/tests/integration/management/test_key_delete_revocation.py @@ -171,11 +171,8 @@ def test_deleting_a_key_revokes_it(gateway: Gateway, spelling: str) -> None: assert deleted_rows == (_DeletedKeyRow(token=key_hash),) -def test_deleting_a_key_revokes_it_on_a_warmed_peer(gateway: Gateway, peer: Gateway) -> None: - pytest.skip( - "BUG: POST /key/delete does not broadcast the key eviction, so a warmed peer keeps answering 200 for about " - "60 seconds until its in-memory cache TTL expires" - ) +@pytest.mark.parametrize("spelling", ("keys", "hashed_keys", "key_aliases")) +def test_deleting_a_key_revokes_it_on_a_warmed_peer(gateway: Gateway, peer: Gateway, spelling: str) -> None: backend: Final = "integration-keys-auth-delete" prompts: Final = ( "delete warm primary " + uuid.uuid4().hex, @@ -208,9 +205,13 @@ def test_deleting_a_key_revokes_it_on_a_warmed_peer(gateway: Gateway, peer: Gate ) assert len(warm_responses) == 2 assert tuple((request.method, request.target) for request in wire.drain()) == (_EXPECTED_CHAT,) * 2 - deleted: Final = gateway.request("POST", "/key/delete", {"keys": [key]}) + identifier: Final = {"keys": key, "hashed_keys": sha256(key.encode()).hexdigest(), "key_aliases": alias}[ + spelling + ] + delete_body: Final = {"key_aliases" if spelling == "key_aliases" else "keys": [identifier]} + deleted: Final = gateway.request("POST", "/key/delete", delete_body) assert deleted.status_code == 200, deleted.text - assert _DeleteResponse.model_validate_json(deleted.content).deleted_keys == (key,), deleted.text + assert _DeleteResponse.model_validate_json(deleted.content).deleted_keys == (identifier,), deleted.text peer_refusal: Final = eventually( lambda: _chat(peer, model, key, prompts[2]), lambda response: response.status_code == 401, diff --git a/tests/unit/proxy/auth/test_auth_checks.py b/tests/unit/proxy/auth/test_auth_checks.py index e783f01769e..9d4fe67614e 100644 --- a/tests/unit/proxy/auth/test_auth_checks.py +++ b/tests/unit/proxy/auth/test_auth_checks.py @@ -46,7 +46,7 @@ from litellm.proxy.auth.auth_checks import ( _team_soft_budget_check, common_checks, ) -from litellm.proxy.utils import ProxyLogging +from litellm.proxy.utils import InternalUsageCache, ProxyLogging from litellm.proxy.utils import CallInfo from litellm.types.passthrough_endpoints.pass_through_endpoints import LITELLM_PASS_THROUGH_ENDPOINT_MARKER @@ -2425,3 +2425,51 @@ async def test_bulk_eviction_clears_the_redis_copies_before_waiting_on_a_wedged_ assert refilled == [] assert redis.entries == {} + + +class _SignallingPublishRedisClient(_SlowPublishRedisClient): + def __init__(self) -> None: + super().__init__() + self.sent = asyncio.Event() + + async def _send(self, messages: Sequence[str]) -> None: + await super()._send(messages) + self.sent.set() + + +class _DeleteAfterBroadcastRedisCache(_AuthRedisCache): + def __init__(self, broadcast_sent: asyncio.Event) -> None: + super().__init__({}) + self._broadcast_sent = broadcast_sent + self.finished = False + + async def delete_cache_keys(self, keys: Sequence[str]) -> None: + await self._broadcast_sent.wait() + self.finished = True + + +@pytest.mark.asyncio +async def test_bulk_eviction_broadcasts_while_the_usage_cache_redis_is_wedged( + monkeypatch: pytest.MonkeyPatch, +) -> None: + monkeypatch.setattr(pubsub_module, "_in_flight_publishes", asyncio.Semaphore(16)) + monkeypatch.setattr(pubsub_module, "_pending_publishes", set()) + client: Final = _SignallingPublishRedisClient() + usage_redis: Final = _DeleteAfterBroadcastRedisCache(broadcast_sent=client.sent) + proxy_logging_obj: Final = ProxyLogging(user_api_key_cache=None) + proxy_logging_obj.internal_usage_cache = InternalUsageCache( + dual_cache=DualCache(redis_cache=usage_redis) # pyright: ignore[reportArgumentType] # a dict-backed fake of the Redis layer + ) + hashed_tokens: Final = ("token-a", "token-b") + + with patch( + "litellm.proxy.common_utils.auth_cache_invalidation_pubsub.coordination_redis_cache", + return_value=_PubSubRedisCache(client), + ): + await delete_cache_key_objects( + hashed_tokens=hashed_tokens, user_api_key_cache=UserApiKeyCache(), proxy_logging_obj=proxy_logging_obj + ) + await _drain_publishes() + + assert usage_redis.finished + assert {json.loads(message)["cache_key"] for message in client.published} == set(hashed_tokens) diff --git a/tests/unit/proxy/management_endpoints/test_key_management_endpoints.py b/tests/unit/proxy/management_endpoints/test_key_management_endpoints.py index 62dd367b8e5..c26a4bf19b8 100644 --- a/tests/unit/proxy/management_endpoints/test_key_management_endpoints.py +++ b/tests/unit/proxy/management_endpoints/test_key_management_endpoints.py @@ -7,7 +7,7 @@ from datetime import datetime, timedelta, timezone from types import SimpleNamespace from typing import Final from unittest.mock import AsyncMock, MagicMock, patch -from uuid import UUID +from uuid import UUID, uuid4 import pytest import yaml @@ -16,6 +16,7 @@ from fastapi.testclient import TestClient from pydantic import JsonValue, ValidationError import litellm +from litellm.caching.in_memory_cache import InMemoryCache from litellm.constants import PROXY_DB_LOOKUP_STALL_WINDOW_SECONDS from litellm.litellm_core_utils.duration_parser import duration_in_seconds from litellm.models.object_permission import LiteLLM_ObjectPermissionTable @@ -41,12 +42,15 @@ from litellm.proxy._types import ( RegenerateKeyRequest, ResetSpendRequest, UpdateKeyRequest, + hash_token, ) from litellm.proxy.auth.auth_checks import ( delete_cache_key_object, + get_key_object, jwt_key_mapping_cache_key, ) from litellm.proxy.auth.user_api_key_auth import UserAPIKeyAuth +from litellm.proxy.common_utils.auth_cache_invalidation_pubsub import AuthCacheInvalidationSubscriber from litellm.proxy.common_utils.user_api_key_cache import UserApiKeyCache, project_cache_key from litellm.proxy.db.db_lookup_gate import db_lookup_stall_tracker from litellm.proxy.management_endpoints.key_management_endpoints import ( @@ -5755,6 +5759,231 @@ async def test_delete_verification_tokens_evicts_jwt_key_mapping_cache(monkeypat ) +class _ChannelSubscription: + """Marks a message done when the subscriber comes back for the next one, so ``messages.join()`` + returns once every delivered message has been applied""" + + def __init__(self) -> None: + self.messages: asyncio.Queue[dict[str, object]] = asyncio.Queue() + self._holding_message = False + + async def subscribe(self, *channels: str) -> None: + return None + + async def get_message(self, *, ignore_subscribe_messages: bool, timeout: float) -> dict[str, object] | None: + if self._holding_message: + self._holding_message = False + self.messages.task_done() + try: + message: Final = await asyncio.wait_for(self.messages.get(), timeout) + except asyncio.TimeoutError: + return None + self._holding_message = True + return message + + async def aclose(self) -> None: + return None + + +class _ChannelPipeline: + def __init__(self, redis: "_SharedCoordinationRedis") -> None: + self._redis = redis + self._messages: tuple[str, ...] = () + + def publish(self, channel: str, message: str) -> "_ChannelPipeline": + self._messages = (*self._messages, message) + return self + + async def execute(self) -> None: + for message in self._messages: + self._redis.deliver(message) + + +class _SharedCoordinationRedis: + """The coordination Redis two workers share: whatever one worker publishes reaches the other's subscriber""" + + namespace = None + + def __init__(self) -> None: + self.published: tuple[str, ...] = () + self._subscriptions: tuple[_ChannelSubscription, ...] = () + + def init_pubsub_client(self) -> "_SharedCoordinationRedis": + return self + + def pubsub(self) -> _ChannelSubscription: + subscription: Final = _ChannelSubscription() + self._subscriptions = (*self._subscriptions, subscription) + return subscription + + def pipeline(self, transaction: bool = True) -> _ChannelPipeline: + return _ChannelPipeline(self) + + async def publish(self, channel: str, message: str) -> int: + self.deliver(message) + return 1 + + def deliver(self, message: str) -> None: + self.published = (*self.published, message) + for subscription in self._subscriptions: + subscription.messages.put_nowait({"type": "message", "data": message.encode()}) + + async def until_every_subscriber_applied_its_messages(self) -> None: + _ = await asyncio.gather(*(subscription.messages.join() for subscription in self._subscriptions)) + + +def _owned_key_row(hashed_token: str) -> LiteLLM_VerificationToken: + return LiteLLM_VerificationToken( + token=hashed_token, + user_id="user-123", + team_id=None, + key_alias="doomed-key", + spend=0.0, + max_budget=None, + models=[], + aliases={}, + config={}, + permissions={}, + metadata={}, + model_max_budget={}, + model_spend={}, + soft_budget_cooldown=False, + allowed_routes=[], + ) + + +def _prisma_with_key_row(key_row: LiteLLM_VerificationToken) -> AsyncMock: + prisma_client: Final = AsyncMock() + prisma_client.db.litellm_verificationtoken.find_many = AsyncMock(return_value=[key_row]) + prisma_client.db.litellm_jwtkeymapping = _CascadingJWTMappingTable([]) + prisma_client.db.litellm_deletedverificationtoken.create_many = AsyncMock() + prisma_client.delete_data = AsyncMock(return_value=[key_row.token]) + prisma_client.get_data = AsyncMock(return_value=None) + return prisma_client + + +@pytest.mark.asyncio +async def test_delete_verification_tokens_stops_the_key_on_a_peer_worker(monkeypatch: pytest.MonkeyPatch) -> None: + """Auth serves key objects cache-first, so a peer worker that is not told about the delete keeps + authenticating the deleted key from its own copy until user_api_key_cache_ttl expires""" + raw_key: Final = "sk-" + uuid4().hex + hashed_token: Final = hash_token(raw_key) + deleting_worker_cache: Final = UserApiKeyCache() + peer_worker_cache: Final = UserApiKeyCache() + for cache in (deleting_worker_cache, peer_worker_cache): + cache.set_cache(key=hashed_token, value=UserAPIKeyAuth(token=hashed_token, user_id="user-123")) + coordination_redis: Final = _SharedCoordinationRedis() + peer_subscriber: Final = AuthCacheInvalidationSubscriber( + redis_cache=coordination_redis, user_api_key_cache=peer_worker_cache + ) + peer_subscriber.start() + prisma_client: Final = _prisma_with_key_row(_owned_key_row(hashed_token)) + monkeypatch.setattr("litellm.proxy.proxy_server.redis_usage_cache", coordination_redis) + monkeypatch.setattr("litellm.proxy.proxy_server.prisma_client", prisma_client) + + try: + await delete_verification_tokens( + tokens=[raw_key], + user_api_key_cache=deleting_worker_cache, + user_api_key_dict=UserAPIKeyAuth( + user_id="admin-user", api_key="sk-admin", user_role=LitellmUserRoles.PROXY_ADMIN.value + ), + ) + await coordination_redis.until_every_subscriber_applied_its_messages() + finally: + await peer_subscriber.stop() + + assert deleting_worker_cache.get_cache(key=hashed_token) is None + with pytest.raises(ProxyException) as denied: + await get_key_object( + hashed_token=hashed_token, prisma_client=prisma_client, user_api_key_cache=peer_worker_cache + ) + assert denied.value.code == "401" + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "deleter", + [ + UserAPIKeyAuth(user_id="admin-user", api_key="sk-admin", user_role=LitellmUserRoles.PROXY_ADMIN.value), + UserAPIKeyAuth(user_id="user-123", api_key="sk-owner", user_role=LitellmUserRoles.INTERNAL_USER.value), + ], + ids=["proxy_admin", "key_owner"], +) +async def test_delete_verification_tokens_evicts_a_requested_key_whose_row_is_already_gone( + monkeypatch: pytest.MonkeyPatch, deleter: UserAPIKeyAuth +) -> None: + """A key whose row was removed outside the API stays cached on every worker, so deleting it in a + batch beside a live key must still stop it everywhere instead of only evicting the rows found""" + live_key: Final = "sk-" + uuid4().hex + gone_key: Final = "sk-" + uuid4().hex + live_token: Final = hash_token(live_key) + gone_token: Final = hash_token(gone_key) + deleting_worker_cache: Final = UserApiKeyCache() + peer_worker_cache: Final = UserApiKeyCache() + for cache in (deleting_worker_cache, peer_worker_cache): + for token in (live_token, gone_token): + cache.set_cache(key=token, value=UserAPIKeyAuth(token=token, user_id="user-123")) + coordination_redis: Final = _SharedCoordinationRedis() + peer_subscriber: Final = AuthCacheInvalidationSubscriber( + redis_cache=coordination_redis, user_api_key_cache=peer_worker_cache + ) + peer_subscriber.start() + monkeypatch.setattr("litellm.proxy.proxy_server.redis_usage_cache", coordination_redis) + monkeypatch.setattr("litellm.proxy.proxy_server.prisma_client", _prisma_with_key_row(_owned_key_row(live_token))) + + try: + await delete_verification_tokens( + tokens=[live_key, gone_key], user_api_key_cache=deleting_worker_cache, user_api_key_dict=deleter + ) + await coordination_redis.until_every_subscriber_applied_its_messages() + finally: + await peer_subscriber.stop() + + assert [deleting_worker_cache.get_cache(key=token) for token in (live_token, gone_token)] == [None, None] + assert [peer_worker_cache.get_cache(key=token) for token in (live_token, gone_token)] == [None, None] + + +@pytest.mark.asyncio +async def test_delete_verification_tokens_leaves_unowned_cache_entries_alone_on_every_worker(monkeypatch: pytest.MonkeyPatch) -> None: + """/key/delete accepts any string, and a name that is not a key row passes through unhashed. A key + owner padding the request with another team's spend counter name must not wipe that counter from + every worker's memory: only the rows the caller was allowed to delete are evicted and broadcast""" + raw_key: Final = "sk-" + uuid4().hex + hashed_token: Final = hash_token(raw_key) + other_team_counter: Final = "spend:team:other-team" + deleting_worker_cache: Final = UserApiKeyCache() + peer_worker_cache: Final = UserApiKeyCache() + peer_spend_counters: Final = InMemoryCache() + peer_spend_counters.set_cache(other_team_counter, 42.0) + peer_worker_cache.set_cache(key=hashed_token, value=UserAPIKeyAuth(token=hashed_token, user_id="user-123")) + coordination_redis: Final = _SharedCoordinationRedis() + peer_subscriber: Final = AuthCacheInvalidationSubscriber( + redis_cache=coordination_redis, + user_api_key_cache=peer_worker_cache, + additional_in_memory_caches=(peer_spend_counters,), + ) + peer_subscriber.start() + monkeypatch.setattr("litellm.proxy.proxy_server.redis_usage_cache", coordination_redis) + monkeypatch.setattr("litellm.proxy.proxy_server.prisma_client", _prisma_with_key_row(_owned_key_row(hashed_token))) + + try: + await delete_verification_tokens( + tokens=[other_team_counter, raw_key], + user_api_key_cache=deleting_worker_cache, + user_api_key_dict=UserAPIKeyAuth( + user_id="user-123", api_key="sk-owner", user_role=LitellmUserRoles.INTERNAL_USER.value + ), + ) + await coordination_redis.until_every_subscriber_applied_its_messages() + finally: + await peer_subscriber.stop() + + assert peer_worker_cache.get_cache(key=hashed_token) is None + assert peer_spend_counters.get_cache(other_team_counter) == 42.0 + assert coordination_redis.published == (json.dumps({"cache_key": hashed_token}),) + + @pytest.mark.asyncio async def test_delete_key_fn_persists_deleted_keys(monkeypatch): from litellm.proxy._types import KeyRequest From 17d1f5833689739003abf379950137d7e8eeb2f5 Mon Sep 17 00:00:00 2001 From: tin-berri Date: Fri, 30 Oct 2026 10:31:57 -0700 Subject: [PATCH 15/18] fix(ui): preserve team model grants on unrelated settings saves (#49258) --- .../ModelSelect/ModelSelect.test.tsx | 33 +++++++++++ .../components/ModelSelect/ModelSelect.tsx | 5 +- .../team/TeamInfo.integration.test.tsx | 59 ++++++++++++++++++- .../src/components/team/TeamInfo.test.tsx | 1 - .../src/components/team/TeamInfo.tsx | 5 +- 5 files changed, 97 insertions(+), 6 deletions(-) diff --git a/ui/litellm-dashboard/src/components/ModelSelect/ModelSelect.test.tsx b/ui/litellm-dashboard/src/components/ModelSelect/ModelSelect.test.tsx index 7ab757cfed9..d7394b1f949 100644 --- a/ui/litellm-dashboard/src/components/ModelSelect/ModelSelect.test.tsx +++ b/ui/litellm-dashboard/src/components/ModelSelect/ModelSelect.test.tsx @@ -775,6 +775,39 @@ describe("ModelSelect", () => { expect(mockOnChange).toHaveBeenCalledWith(["gpt-4"]); }); + it.each([ + { value: "all-proxy-models", label: "All Proxy Models", includeSpecialOptions: true }, + { value: "no-default-models", label: "No Default Models", includeSpecialOptions: false }, + ])( + "should label retained $value without offering it again after removal", + async ({ value, label, includeSpecialOptions }) => { + const user = userEvent.setup(); + mockUseOrganization.mockReturnValue({ + data: createMockOrganization(["gpt-4"]), + isLoading: false, + } as unknown as ReturnType); + const props = { + onChange: mockOnChange, + context: "team" as const, + organizationID: "org-1", + options: { includeSpecialOptions }, + }; + const { rerender } = renderWithProviders(); + + expect(screen.getByLabelText(label)).toBeInTheDocument(); + await openModelList(user); + const retainedGroup = within(screen.getByRole("listbox")).getAllByRole("group")[0]; + expect(within(retainedGroup).getByText("Unavailable")).toBeInTheDocument(); + await user.click(within(retainedGroup).getByRole("option", { name: label })); + expect(mockOnChange).toHaveBeenCalledWith([]); + + rerender(); + expectNotOffered(label); + expectNotOffered("claude-3"); + expectOffered("gpt-4"); + }, + ); + it("should keep the other selections when removing one unavailable model alongside a special option", async () => { const user = userEvent.setup(); renderWithProviders( diff --git a/ui/litellm-dashboard/src/components/ModelSelect/ModelSelect.tsx b/ui/litellm-dashboard/src/components/ModelSelect/ModelSelect.tsx index f169382b4ed..c7c428f2ccf 100644 --- a/ui/litellm-dashboard/src/components/ModelSelect/ModelSelect.tsx +++ b/ui/litellm-dashboard/src/components/ModelSelect/ModelSelect.tsx @@ -240,7 +240,10 @@ export const ModelSelect = (props: ModelSelectProps) => { ]; const offeredByValue = new Map(offeredGroups.flatMap((group) => group.items).map((option) => [option.value, option])); - const selectedOptions = value.map((v) => offeredByValue.get(v) ?? { label: v, value: v }); + const selectedOptions = value.map( + (v) => + offeredByValue.get(v) ?? MODEL_SENTINEL_OPTIONS.find((option) => option.value === v) ?? { label: v, value: v }, + ); const groups: ModelOptionGroup[] = [ ...unavailableGroups( selectedOptions, diff --git a/ui/litellm-dashboard/src/components/team/TeamInfo.integration.test.tsx b/ui/litellm-dashboard/src/components/team/TeamInfo.integration.test.tsx index fc054e51e5b..545c504281b 100644 --- a/ui/litellm-dashboard/src/components/team/TeamInfo.integration.test.tsx +++ b/ui/litellm-dashboard/src/components/team/TeamInfo.integration.test.tsx @@ -1,6 +1,6 @@ import { useTeamMetadataSchema } from "@/app/(dashboard)/hooks/teams/useTeamMetadataSchema"; import * as networking from "@/components/networking"; -import { fireEvent, screen, waitFor } from "@testing-library/react"; +import { fireEvent, screen, waitFor, within } from "@testing-library/react"; import userEvent from "@testing-library/user-event"; import { afterEach, beforeEach, describe, expect, it, vi } from "vitest"; import { renderWithProviders } from "../../../tests/test-utils"; @@ -817,7 +817,7 @@ describe("TeamInfoView - member budget reset prompt", () => { }); }); -describe("TeamInfoView - model rate limit picker", () => { +describe("TeamInfoView - model settings", () => { const props = { teamId: "123", onUpdate: vi.fn(), @@ -836,6 +836,61 @@ describe("TeamInfoView - model rate limit picker", () => { vi.clearAllMocks(); }); + it.each([ + { name: "unrestricted", models: [], organizationId: null }, + { name: "explicit all models", models: ["all-proxy-models"], organizationId: null }, + { name: "access groups only", models: ["no-default-models"], organizationId: null }, + { name: "retained explicit models", models: ["retired-model", "openai/*"], organizationId: null }, + { name: "unrestricted within an organization", models: [], organizationId: "org-1" }, + ])("preserves $name model grants when saving only a budget", async ({ models, organizationId }) => { + const user = userEvent.setup({ delay: null }); + vi.mocked(networking.teamInfoCall).mockResolvedValue( + createMockTeamData({ models, organization_id: organizationId }), + ); + mockUseTeam.mockReturnValue({ + data: { team_id: "123", organization_models: organizationId ? ["gpt-4"] : undefined }, + isLoading: false, + } as unknown as ReturnType); + + renderWithProviders(); + await user.click(await screen.findByRole("tab", { name: "Settings" })); + await user.click(await screen.findByRole("button", { name: /edit settings/i })); + if (models.length === 0) { + expect(await screen.findByLabelText("All Proxy Models")).toBeInTheDocument(); + } + fireEvent.change(await screen.findByLabelText("Max Budget (USD)"), { target: { value: "100" } }); + await user.click(screen.getByRole("button", { name: /save changes/i })); + + await waitFor(() => expect(networking.teamUpdateCall).toHaveBeenCalledTimes(1)); + const payload = vi.mocked(networking.teamUpdateCall).mock.calls[0][1]; + expect(payload).toMatchObject({ team_id: "123", max_budget: 100 }); + expect(payload).not.toHaveProperty("models"); + }); + + it("discards a cancelled model clear but saves an intentional clear of unrestricted access", async () => { + const user = userEvent.setup({ delay: null }); + vi.mocked(networking.teamInfoCall).mockResolvedValue(createMockTeamData({ models: [] })); + + renderWithProviders(); + await user.click(await screen.findByRole("tab", { name: "Settings" })); + await user.click(await screen.findByRole("button", { name: /edit settings/i })); + await user.click(within(await screen.findByLabelText("All Proxy Models")).getByRole("button")); + await user.click(screen.getByRole("button", { name: "Cancel" })); + await user.click(await screen.findByRole("button", { name: /edit settings/i })); + expect(await screen.findByLabelText("All Proxy Models")).toBeInTheDocument(); + fireEvent.change(screen.getByLabelText("Max Budget (USD)"), { target: { value: "100" } }); + await user.click(screen.getByRole("button", { name: /save changes/i })); + + await waitFor(() => expect(networking.teamUpdateCall).toHaveBeenCalledTimes(1)); + expect(vi.mocked(networking.teamUpdateCall).mock.calls[0][1]).not.toHaveProperty("models"); + await user.click(await screen.findByRole("button", { name: /edit settings/i })); + await user.click(within(await screen.findByLabelText("All Proxy Models")).getByRole("button")); + await user.click(screen.getByRole("button", { name: /save changes/i })); + + await waitFor(() => expect(networking.teamUpdateCall).toHaveBeenCalledTimes(2)); + expect(vi.mocked(networking.teamUpdateCall).mock.calls[1][1].models).toEqual(["no-default-models"]); + }); + it("offers access-group models, not no-default-models, in the model rate limit picker", async () => { const user = userEvent.setup({ delay: null }); vi.mocked(networking.teamInfoCall).mockResolvedValue( diff --git a/ui/litellm-dashboard/src/components/team/TeamInfo.test.tsx b/ui/litellm-dashboard/src/components/team/TeamInfo.test.tsx index e4bf67ab72f..46081ae96f5 100644 --- a/ui/litellm-dashboard/src/components/team/TeamInfo.test.tsx +++ b/ui/litellm-dashboard/src/components/team/TeamInfo.test.tsx @@ -2421,7 +2421,6 @@ describe("TeamInfoView - the exact bytes the update call sends", () => { const alwaysSent = { team_id: "123", team_alias: "Test Team", - models: ["gpt-4"], tpm_limit: 1000, rpm_limit: 1000, tpd_limit: null, diff --git a/ui/litellm-dashboard/src/components/team/TeamInfo.tsx b/ui/litellm-dashboard/src/components/team/TeamInfo.tsx index 0d20d160283..329552d735a 100644 --- a/ui/litellm-dashboard/src/components/team/TeamInfo.tsx +++ b/ui/litellm-dashboard/src/components/team/TeamInfo.tsx @@ -72,6 +72,7 @@ import { } from "../key_team_helpers/ModelMaxBudgetEditor"; import { modelMaxBudgetUpdate, StoredModelMaxBudget } from "../key_team_helpers/modelMaxBudgetPayload"; import { + ALL_PROXY_MODELS, computeTeamModelBadges, normalizeTeamModelSelection, TeamAccessGroupModelGrant, @@ -515,7 +516,7 @@ const computeEffectiveGuardrails = (info: TeamInfoRecord, globalGuardrailNames: const toTeamFormValues = (info: TeamInfoRecord, effectiveGuardrails: string[]): TeamUpdateFormValues => ({ team_alias: info.team_alias, - models: info.models, + models: info.models.length === 0 ? [ALL_PROXY_MODELS] : info.models, max_budget: info.max_budget, soft_budget: info.soft_budget, soft_budget_alerting_emails: Array.isArray(info.metadata?.soft_budget_alerting_emails) @@ -1064,7 +1065,7 @@ const TeamInfoView: React.FC = ({ const updateData: any = { team_id: teamId, team_alias: values.team_alias, - models: normalizeTeamModelSelection(values.models), + ...(form.getFieldState("models").isDirty ? { models: normalizeTeamModelSelection(values.models) } : {}), tpm_limit: numberOrNull(values.tpm_limit), rpm_limit: numberOrNull(values.rpm_limit), tpd_limit: numberOrNull(values.tpd_limit), From 707ac1ffee620e2906ef0fb5f9ad699cfd65a16d Mon Sep 17 00:00:00 2001 From: islomjon-aiand Date: Sat, 31 Oct 2026 02:55:26 +0900 Subject: [PATCH 16/18] fix(aiand)!: show ai&'s name and logo, and sync ai&'s model list (#49158) * feat(ui): show ai&'s name and logo in the dashboard The Add Model form already lists ai& from its provider metadata, but the dashboard had no entry for it, so the provider showed no logo and model tables printed the raw aiand slug. Adds ai& to the dashboard provider list with its official mark, plated in dark mode so the red and black mark keep their colours, and the aiand/openai/gpt-oss-120b model placeholder * fix(aiand): retire deepseek-v4-flash, glm-5.2 and kimi-k2.7-code ai& retired these three models on 2026-10-27 and https://api.aiand.com/v1/models now lists ten. The rows keep their prices with a deprecation_date, so earlier spend still prices, and aiand/* stops listing them, the same way baseten handles retired rows. The other ten rows already match the live prices, context windows and output limits * feat(aiand): add deepseek-v4.1-flash-1017 ai& launches deepseek-ai/deepseek-v4.1-flash-1017 on 2026-11-02. It is served on https://staging-api.aiand.com/v1/models with the same price, context window, output limit and capabilities as deepseek-v4.1-flash, so its row copies that one * fix(aiand): retire deepseek-v4.1-flash ai& retires deepseek-ai/deepseek-v4.1-flash on 2026-10-30, now that deepseek-v4.1-flash-1017 is live in production on https://api.aiand.com/v1/models with the same price, limits and capabilities. Its row keeps its prices with a deprecation_date, like the other three retired ai& rows * fix(aiand): retire deepseek-v4.1-flash on 2026-11-06 https://api.aiand.com/v1/models now gives deepseek-v4.1-flash a deprecation_date of 2026-11-05T15:00Z, which is 2026-11-06 00:00 in Japan, so its row uses 2026-11-06 instead of 2026-10-30 and aiand/* keeps listing it until then. The wildcard test now leaves a model out only once its retirement date has passed * test(aiand): derive retired models from the rows instead of pinning dates The wildcard test now reads each row's deprecation_date, so it holds whichever dates ai& sets, and the test that only restated the JSON dates is gone * fix(aiand): retire models on their UTC date while the proxy runs, and add muse-spark-open The wildcard listing now checks each row's deprecation_date when it expands, against the UTC date the field documents, so a model drops on its day without a restart. deepseek-v4.1-flash now carries 2026-11-05, the UTC day ai& retires it, and meta/muse-spark-open gets its price row from ai&'s live /v1/models * fix(aiand): hide retired models only from bundled listings, including the wildcard health check * fix(aiand): share the UTC retirement check with Together AI and the health check sort --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> --- litellm/__init__.py | 21 ++-- .../litellm_core_utils/llm_request_utils.py | 21 ++-- .../litellm_core_utils/model_retirement.py | 19 ++++ ...odel_prices_and_context_window_backup.json | 34 ++++++ litellm/utils.py | 10 +- model_prices_and_context_window.json | 34 ++++++ .../test_llm_request_utils.py | 38 +++++++ .../baseten/chat/test_baseten_completions.py | 33 +++++- .../llms/openai_like/test_aiand_provider.py | 100 +++++++++++++++++- .../public/assets/logos/aiand.svg | 4 + .../components/provider_info_helpers.test.tsx | 12 +++ .../src/components/provider_info_helpers.tsx | 5 + .../src/lib/logoTreatments.ts | 1 + 13 files changed, 296 insertions(+), 36 deletions(-) create mode 100644 litellm/litellm_core_utils/model_retirement.py create mode 100644 ui/litellm-dashboard/public/assets/logos/aiand.svg diff --git a/litellm/__init__.py b/litellm/__init__.py index 5f14c605cd0..f64531f86fb 100644 --- a/litellm/__init__.py +++ b/litellm/__init__.py @@ -7,7 +7,6 @@ _apply_process_defaults() ### INIT VARIABLES ######################### import threading import os -import datetime from collections.abc import Mapping, Sequence from typing import ( @@ -550,6 +549,7 @@ _key_management_system: Optional["KeyManagementSystem"] = None output_parse_pii: bool = False ############################################# from litellm.litellm_core_utils.get_model_cost_map import get_model_cost_map, mark_litellm_import_complete +from litellm.litellm_core_utils.model_retirement import is_hidden_as_retired model_cost = get_model_cost_map(url=model_cost_map_url) cost_discount_config: Dict[str, float] = {} # Provider-specific cost discounts {"vertex_ai": 0.05} = 5% discount @@ -756,16 +756,6 @@ def is_openai_finetune_model(key: str) -> bool: return key.startswith("ft:") and not key.count(":") > 1 -def _is_past_deprecation_date(value: Mapping[str, object]) -> bool: - deprecation_date: Final = value.get("deprecation_date") - if not isinstance(deprecation_date, str): - return False - try: - return datetime.date.fromisoformat(deprecation_date) <= datetime.date.today() - except ValueError: - return False - - def _populate_provider_model_sets(model_cost_map: Mapping[str, Mapping[str, object]]) -> None: for key, value in tuple(model_cost_map.items()): if value.get("litellm_provider") == "openai" and not is_openai_finetune_model(key): @@ -873,12 +863,12 @@ def _populate_provider_model_sets(model_cost_map: Mapping[str, Mapping[str, obje if "-to-" not in key and "fireworks-ai-default" not in key: fireworks_ai_models.add(key) elif value.get("litellm_provider") == "together_ai": - if _is_past_deprecation_date(value): + if is_hidden_as_retired(value): together_ai_models.discard(key) elif not key.startswith("together-ai-"): together_ai_models.add(key) elif value.get("litellm_provider") == "baseten": - if _is_past_deprecation_date(value): + if is_hidden_as_retired(value): baseten_models.discard(key) else: baseten_models.add(key) @@ -1051,7 +1041,10 @@ def _populate_provider_model_sets(model_cost_map: Mapping[str, Mapping[str, obje elif value.get("litellm_provider") == "coralbricks": coralbricks_models.add(key) elif value.get("litellm_provider") == "aiand": - aiand_models.add(key) + if is_hidden_as_retired(value): + aiand_models.discard(key) + else: + aiand_models.add(key) elif value.get("litellm_provider") == "reducto": reducto_models.add(key) elif value.get("litellm_provider") == "bedrock_mantle": diff --git a/litellm/litellm_core_utils/llm_request_utils.py b/litellm/litellm_core_utils/llm_request_utils.py index 9c2ec0b8984..84f161e0c62 100644 --- a/litellm/litellm_core_utils/llm_request_utils.py +++ b/litellm/litellm_core_utils/llm_request_utils.py @@ -1,4 +1,3 @@ -import datetime from collections.abc import Iterable, Iterator, Mapping from dataclasses import dataclass from typing import Final, cast @@ -9,6 +8,7 @@ from typing_extensions import NotRequired, ReadOnly, TypedDict import litellm from litellm.constants import DEFAULT_MAX_RECURSE_DEPTH from litellm.litellm_core_utils.get_llm_provider_logic import inferred_provider +from litellm.litellm_core_utils.model_retirement import is_past_deprecation_date def _form_field_value(value: object) -> str: @@ -168,7 +168,9 @@ class RankedChatModel: def rank_chat_models_from_llm_provider(custom_llm_provider: str) -> tuple[RankedChatModel, ...]: - known_models: Final[Iterable[str]] = litellm.models_by_provider.get(custom_llm_provider, ()) + from litellm.utils import bundled_provider_models + + known_models: Final = bundled_provider_models(custom_llm_provider) ranked_models: Final = ( RankedChatModel( model=model, route_rank=route_rank, cost=cost, is_past_deprecation_date=is_past_deprecation_date @@ -187,18 +189,11 @@ _COST_MAP_DEPRECATION: Final = TypeAdapter(_CostMapDeprecation) def _is_past_deprecation_date(cost_map_key: str) -> bool: - deprecation_date: Final = _COST_MAP_DEPRECATION.validate_python( - litellm.model_cost.get(cost_map_key) or {} # pyright: ignore[reportUnknownMemberType] # untyped cost map JSON - ).get("deprecation_date") - if not isinstance(deprecation_date, str): - return False - try: - return ( - datetime.date.fromisoformat(deprecation_date) - <= datetime.datetime.now(datetime.timezone.utc).astimezone().date() + return is_past_deprecation_date( + _COST_MAP_DEPRECATION.validate_python( + litellm.model_cost.get(cost_map_key) or {} # pyright: ignore[reportUnknownMemberType] # untyped cost map JSON ) - except ValueError: - return False + ) def _chat_row_costs(models: Iterable[str], custom_llm_provider: str) -> Iterator[tuple[str, float, bool]]: diff --git a/litellm/litellm_core_utils/model_retirement.py b/litellm/litellm_core_utils/model_retirement.py new file mode 100644 index 00000000000..3bfb834733d --- /dev/null +++ b/litellm/litellm_core_utils/model_retirement.py @@ -0,0 +1,19 @@ +import datetime +from collections.abc import Mapping +from typing import Final + +PROVIDERS_HIDING_RETIRED_MODELS: Final = frozenset({"aiand", "baseten", "together_ai"}) + + +def is_past_deprecation_date(value: Mapping[str, object]) -> bool: + deprecation_date: Final = value.get("deprecation_date") + if not isinstance(deprecation_date, str): + return False + try: + return datetime.date.fromisoformat(deprecation_date) <= datetime.datetime.now(datetime.timezone.utc).date() + except ValueError: + return False + + +def is_hidden_as_retired(value: Mapping[str, object]) -> bool: + return value.get("litellm_provider") in PROVIDERS_HIDING_RETIRED_MODELS and is_past_deprecation_date(value) diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json index e1131117fcf..ea9cacbdabe 100644 --- a/litellm/model_prices_and_context_window_backup.json +++ b/litellm/model_prices_and_context_window_backup.json @@ -72249,6 +72249,7 @@ }, "aiand/deepseek-ai/deepseek-v4-flash": { "cache_read_input_token_cost": 8e-08, + "deprecation_date": "2026-10-27", "input_cost_per_token": 1.5e-07, "litellm_provider": "aiand", "max_input_tokens": 1048576, @@ -72278,6 +72279,22 @@ "supports_vision": false }, "aiand/deepseek-ai/deepseek-v4.1-flash": { + "cache_read_input_token_cost": 2e-08, + "deprecation_date": "2026-11-05", + "input_cost_per_token": 3e-07, + "litellm_provider": "aiand", + "max_input_tokens": 1048576, + "max_output_tokens": 384000, + "mode": "chat", + "output_cost_per_token": 6e-07, + "source": "https://api.aiand.com/v1/models", + "supports_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "supports_vision": true + }, + "aiand/deepseek-ai/deepseek-v4.1-flash-1017": { "cache_read_input_token_cost": 2e-08, "input_cost_per_token": 3e-07, "litellm_provider": "aiand", @@ -72307,8 +72324,24 @@ "supports_tool_choice": true, "supports_vision": true }, + "aiand/meta/muse-spark-open": { + "cache_read_input_token_cost": 1.5e-07, + "input_cost_per_token": 1.25e-06, + "litellm_provider": "aiand", + "max_input_tokens": 1048576, + "max_output_tokens": 943718, + "mode": "chat", + "output_cost_per_token": 4.25e-06, + "source": "https://api.aiand.com/v1/models", + "supports_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "supports_vision": true + }, "aiand/moonshotai/kimi-k2.7-code": { "cache_read_input_token_cost": 2e-07, + "deprecation_date": "2026-10-27", "input_cost_per_token": 7.5e-07, "litellm_provider": "aiand", "max_input_tokens": 262144, @@ -72399,6 +72432,7 @@ }, "aiand/zai-org/glm-5.2": { "cache_read_input_token_cost": 3e-07, + "deprecation_date": "2026-10-27", "input_cost_per_token": 1e-06, "litellm_provider": "aiand", "max_input_tokens": 1048576, diff --git a/litellm/utils.py b/litellm/utils.py index 6f56cc65f45..2ca8a49f968 100644 --- a/litellm/utils.py +++ b/litellm/utils.py @@ -94,6 +94,7 @@ from litellm.litellm_core_utils.fallback_generalizations import ( ) from litellm.litellm_core_utils.get_model_cost_map import AliasedCostEntry, GetModelCostMap from litellm.litellm_core_utils.model_param_helper import peek_reasoning_summary_aliases +from litellm.litellm_core_utils.model_retirement import is_hidden_as_retired from litellm.litellm_core_utils.offline_mode import is_offline_mode from litellm.litellm_core_utils.sensitive_data_masker import redact_credentials_in_payload from litellm.litellm_core_utils.tokenizer import strip_special_tokens @@ -8806,6 +8807,12 @@ def _get_valid_models_from_provider_api( return [] +def bundled_provider_models(provider: str) -> tuple[str, ...]: + bundled: Final[tuple[str, ...]] = tuple(litellm.models_by_provider.get(provider, ())) + cost_map: Final = _builtin_cost_map() + return tuple(model for model in bundled if not is_hidden_as_retired(cost_map.get(model, {}))) + + def get_valid_models( check_provider_endpoint: bool | None = None, custom_llm_provider: str | None = None, @@ -8871,8 +8878,7 @@ def get_valid_models( ) ) else: - models_for_provider = copy.deepcopy(litellm.models_by_provider.get(provider, [])) - valid_models.extend(models_for_provider) + valid_models.extend(bundled_provider_models(provider)) return valid_models except Exception as e: diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json index e1131117fcf..ea9cacbdabe 100644 --- a/model_prices_and_context_window.json +++ b/model_prices_and_context_window.json @@ -72249,6 +72249,7 @@ }, "aiand/deepseek-ai/deepseek-v4-flash": { "cache_read_input_token_cost": 8e-08, + "deprecation_date": "2026-10-27", "input_cost_per_token": 1.5e-07, "litellm_provider": "aiand", "max_input_tokens": 1048576, @@ -72278,6 +72279,22 @@ "supports_vision": false }, "aiand/deepseek-ai/deepseek-v4.1-flash": { + "cache_read_input_token_cost": 2e-08, + "deprecation_date": "2026-11-05", + "input_cost_per_token": 3e-07, + "litellm_provider": "aiand", + "max_input_tokens": 1048576, + "max_output_tokens": 384000, + "mode": "chat", + "output_cost_per_token": 6e-07, + "source": "https://api.aiand.com/v1/models", + "supports_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "supports_vision": true + }, + "aiand/deepseek-ai/deepseek-v4.1-flash-1017": { "cache_read_input_token_cost": 2e-08, "input_cost_per_token": 3e-07, "litellm_provider": "aiand", @@ -72307,8 +72324,24 @@ "supports_tool_choice": true, "supports_vision": true }, + "aiand/meta/muse-spark-open": { + "cache_read_input_token_cost": 1.5e-07, + "input_cost_per_token": 1.25e-06, + "litellm_provider": "aiand", + "max_input_tokens": 1048576, + "max_output_tokens": 943718, + "mode": "chat", + "output_cost_per_token": 4.25e-06, + "source": "https://api.aiand.com/v1/models", + "supports_function_calling": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_tool_choice": true, + "supports_vision": true + }, "aiand/moonshotai/kimi-k2.7-code": { "cache_read_input_token_cost": 2e-07, + "deprecation_date": "2026-10-27", "input_cost_per_token": 7.5e-07, "litellm_provider": "aiand", "max_input_tokens": 262144, @@ -72399,6 +72432,7 @@ }, "aiand/zai-org/glm-5.2": { "cache_read_input_token_cost": 3e-07, + "deprecation_date": "2026-10-27", "input_cost_per_token": 1e-06, "litellm_provider": "aiand", "max_input_tokens": 1048576, diff --git a/tests/unit/litellm_core_utils/test_llm_request_utils.py b/tests/unit/litellm_core_utils/test_llm_request_utils.py index e7350fae395..5243d308543 100644 --- a/tests/unit/litellm_core_utils/test_llm_request_utils.py +++ b/tests/unit/litellm_core_utils/test_llm_request_utils.py @@ -1,3 +1,5 @@ +import datetime +import time from typing import Final import httpx @@ -245,3 +247,39 @@ def test_rank_chat_models_flags_a_retired_model_the_provider_lists_without_its_c (ranked_model.model, ranked_model.is_past_deprecation_date) for ranked_model in rank_chat_models_from_llm_provider("vertex_ai") ] == [("fake-retired-vertex-chat", True), ("fake-live-vertex-chat", False)] + + +@pytest.mark.parametrize("host_zone", ["Etc/GMT+12", "Etc/GMT-14"]) +def test_rank_chat_models_flags_retirement_by_the_utc_date_whatever_the_host_zone( + monkeypatch: pytest.MonkeyPatch, host_zone: str +) -> None: + utc_today: Final = datetime.datetime.now(datetime.timezone.utc).date() + retirement_dates: Final = { + "deepseek/fake-retires-today-chat": utc_today, + "deepseek/fake-retires-tomorrow-chat": utc_today + datetime.timedelta(days=1), + } + for name, retirement_date in retirement_dates.items(): + monkeypatch.setitem( + litellm.model_cost, + name, + { + "input_cost_per_token": 1e-12, + "output_cost_per_token": 1e-12, + "mode": "chat", + "litellm_provider": PICKER_PROVIDER, + "deprecation_date": retirement_date.isoformat(), + }, + ) + monkeypatch.setitem(litellm.models_by_provider, PICKER_PROVIDER, tuple(retirement_dates)) + monkeypatch.setenv("TZ", host_zone) + time.tzset() + try: + flags: Final = { + ranked_model.model: ranked_model.is_past_deprecation_date + for ranked_model in rank_chat_models_from_llm_provider(PICKER_PROVIDER) + } + finally: + monkeypatch.undo() + time.tzset() + + assert flags == {"deepseek/fake-retires-today-chat": True, "deepseek/fake-retires-tomorrow-chat": False} diff --git a/tests/unit/llms/baseten/chat/test_baseten_completions.py b/tests/unit/llms/baseten/chat/test_baseten_completions.py index 4f135a1a106..407b39319e9 100644 --- a/tests/unit/llms/baseten/chat/test_baseten_completions.py +++ b/tests/unit/llms/baseten/chat/test_baseten_completions.py @@ -7,8 +7,9 @@ import pytest import respx import litellm +from litellm.litellm_core_utils.model_retirement import is_past_deprecation_date from litellm.llms.baseten.chat import BasetenConfig -from litellm.utils import get_valid_models +from litellm.utils import _model_cache, get_valid_models class TestBasetenRouting: @@ -65,7 +66,7 @@ class TestBasetenModelDiscovery: servable_registry_models: Final = { key for key, value in litellm.model_cost.items() - if value.get("litellm_provider") == "baseten" and not litellm._is_past_deprecation_date(value) + if value.get("litellm_provider") == "baseten" and not is_past_deprecation_date(value) } assert servable_registry_models @@ -92,6 +93,20 @@ class TestBasetenModelDiscovery: litellm.baseten_models.discard(model) litellm.add_known_models({}) + def test_wildcard_drops_a_model_whose_retirement_date_passes_while_running(self, monkeypatch: pytest.MonkeyPatch): + model: Final = "baseten/acme/retires-while-running" + monkeypatch.setitem(litellm.model_cost, model, {"litellm_provider": "baseten", "deprecation_date": "2999-01-01"}) + try: + litellm.add_known_models({model: litellm.model_cost[model]}) + assert model in get_valid_models(custom_llm_provider="baseten") + + monkeypatch.setitem(litellm.model_cost, model, {"litellm_provider": "baseten", "deprecation_date": "2000-01-01"}) + + assert model not in get_valid_models(custom_llm_provider="baseten") + finally: + litellm.baseten_models.discard(model) + litellm.add_known_models({}) + @pytest.mark.parametrize( "value, expected", [ @@ -102,7 +117,7 @@ class TestBasetenModelDiscovery: ], ) def test_is_past_deprecation_date(self, value, expected): - assert litellm._is_past_deprecation_date(value) is expected + assert is_past_deprecation_date(value) is expected def test_get_models_calls_model_api(self, respx_mock: respx.MockRouter) -> None: route: Final = respx_mock.get("https://inference.baseten.co/v1/models").mock( @@ -156,6 +171,18 @@ class TestBasetenModelDiscovery: assert models == ["baseten/brand-new/model"] + @patch.dict(os.environ, {"BASETEN_API_KEY": "test-key"}) + def test_live_catalog_keeps_a_model_the_cost_map_dates_as_retired(self, respx_mock: respx.MockRouter) -> None: + assert is_past_deprecation_date(litellm.model_cost["baseten/zai-org/GLM-5"]) + _model_cache.flush_cache() + respx_mock.get("https://inference.baseten.co/v1/models").mock( + return_value=httpx.Response(200, json={"data": [{"id": "zai-org/GLM-5"}]}) + ) + + models: Final = get_valid_models(check_provider_endpoint=True, custom_llm_provider="baseten") + + assert models == ["baseten/zai-org/GLM-5"] + class TestBasetenReasoningEffort: def test_reasoning_effort_passthrough_with_drop_params(self): diff --git a/tests/unit/llms/openai_like/test_aiand_provider.py b/tests/unit/llms/openai_like/test_aiand_provider.py index 804f3a3b6eb..0274b58f19a 100644 --- a/tests/unit/llms/openai_like/test_aiand_provider.py +++ b/tests/unit/llms/openai_like/test_aiand_provider.py @@ -1,4 +1,6 @@ +import datetime import json +import time from typing import Final import httpx @@ -8,6 +10,8 @@ import respx import litellm from litellm.caching.llm_caching_handler import LLMClientCache from litellm.litellm_core_utils.get_llm_provider_logic import get_llm_provider +from litellm.litellm_core_utils.health_check_helpers import wildcard_health_check_models +from litellm.litellm_core_utils.model_retirement import is_past_deprecation_date from litellm.llms.openai_like.json_loader import JSONProviderRegistry from litellm.proxy.auth.model_checks import get_known_models_from_wildcard from litellm.utils import ProviderConfigManager @@ -89,11 +93,99 @@ def _aiand_catalog() -> frozenset[str]: return frozenset(name for name, info in litellm.model_cost.items() if info.get("litellm_provider") == "aiand") -def test_aiand_wildcard_lists_its_priced_models(): - catalog: Final = _aiand_catalog() +def _utc_today() -> datetime.date: + return datetime.datetime.now(datetime.timezone.utc).date() - assert catalog - assert frozenset(get_known_models_from_wildcard("aiand/*")) == catalog + +def test_aiand_wildcard_lists_only_models_not_yet_retired(): + catalog: Final = _aiand_catalog() + retired: Final = frozenset( + name + for name in catalog + if (retired_on := litellm.model_cost[name].get("deprecation_date")) is not None + and datetime.date.fromisoformat(retired_on) <= _utc_today() + ) + + assert catalog - retired + assert frozenset(get_known_models_from_wildcard("aiand/*")) == catalog - retired + + +def test_aiand_wildcard_drops_a_model_whose_retirement_date_passes_while_running(monkeypatch: pytest.MonkeyPatch): + model: Final = "aiand/acme/retires-while-running" + monkeypatch.setitem(litellm.model_cost, model, {"litellm_provider": "aiand", "deprecation_date": "2999-01-01"}) + try: + litellm.add_known_models({model: litellm.model_cost[model]}) + assert model in get_known_models_from_wildcard("aiand/*") + + monkeypatch.setitem(litellm.model_cost, model, {"litellm_provider": "aiand", "deprecation_date": "2000-01-01"}) + + assert model not in get_known_models_from_wildcard("aiand/*") + finally: + litellm.aiand_models.discard(model) + litellm.add_known_models({}) + + +def test_aiand_models_by_provider_leaves_out_models_retired_at_load(): + retired: Final = frozenset( + name + for name, row in litellm.model_cost.items() + if row.get("litellm_provider") == "aiand" and is_past_deprecation_date(row) + ) + + assert retired + assert retired.isdisjoint(litellm.models_by_provider["aiand"]) + + +def test_aiand_wildcard_health_check_skips_a_model_whose_retirement_date_passes_while_running( + monkeypatch: pytest.MonkeyPatch, +): + model: Final = "aiand/acme/free-and-retiring" + live_row: Final = { + "litellm_provider": "aiand", + "mode": "chat", + "input_cost_per_token": 0.0, + "output_cost_per_token": 0.0, + "deprecation_date": "2999-01-01", + } + monkeypatch.setitem(litellm.model_cost, model, live_row) + try: + litellm.add_known_models({model: live_row}) + assert model in wildcard_health_check_models(wildcard="aiand/*", custom_llm_provider="aiand") + + monkeypatch.setitem(litellm.model_cost, model, {**live_row, "deprecation_date": "2000-01-01"}) + + assert model not in wildcard_health_check_models(wildcard="aiand/*", custom_llm_provider="aiand") + finally: + litellm.aiand_models.discard(model) + litellm.add_known_models({}) + + +@pytest.mark.parametrize("host_zone", ["Etc/GMT+12", "Etc/GMT-14"]) +def test_aiand_models_retire_on_the_utc_date_whatever_the_host_zone(monkeypatch: pytest.MonkeyPatch, host_zone: str): + today: Final = _utc_today() + rows: Final = { + "aiand/acme/retires-today": {"litellm_provider": "aiand", "deprecation_date": today.isoformat()}, + "aiand/acme/retires-tomorrow": { + "litellm_provider": "aiand", + "deprecation_date": (today + datetime.timedelta(days=1)).isoformat(), + }, + } + for name, row in rows.items(): + monkeypatch.setitem(litellm.model_cost, name, row) + monkeypatch.setenv("TZ", host_zone) + time.tzset() + try: + litellm.add_known_models(rows) + listed: Final = frozenset(get_known_models_from_wildcard("aiand/*")) + + assert "aiand/acme/retires-today" not in listed + assert "aiand/acme/retires-tomorrow" in listed + finally: + monkeypatch.undo() + time.tzset() + for name in rows: + litellm.aiand_models.discard(name) + litellm.add_known_models({}) def test_aiand_calls_are_billed_from_their_cost_rows(): diff --git a/ui/litellm-dashboard/public/assets/logos/aiand.svg b/ui/litellm-dashboard/public/assets/logos/aiand.svg new file mode 100644 index 00000000000..486c73b8dbc --- /dev/null +++ b/ui/litellm-dashboard/public/assets/logos/aiand.svg @@ -0,0 +1,4 @@ + + + + diff --git a/ui/litellm-dashboard/src/components/provider_info_helpers.test.tsx b/ui/litellm-dashboard/src/components/provider_info_helpers.test.tsx index 810b010f25e..02c31ecc082 100644 --- a/ui/litellm-dashboard/src/components/provider_info_helpers.test.tsx +++ b/ui/litellm-dashboard/src/components/provider_info_helpers.test.tsx @@ -32,6 +32,18 @@ describe("provider_info_helpers", () => { expect(getPlaceholder(Providers.BIGMODEL)).toMatch(/^bigmodel\//); }); + it("resolves ai& metadata, name, logo and model placeholder", () => { + const metadata = PROVIDER_CREATE_FIELDS.find((provider) => provider.provider === "AIAND"); + expect(provider_map.AIAND).toBe(metadata?.litellm_provider); + expect(getProviderLogoAndName("aiand")).toEqual({ + displayName: "ai&", + logo: providerLogoMap[Providers.AIAND], + }); + expect(getProviderLogoAndName("AIAND")).toEqual(getProviderLogoAndName("aiand")); + expect(providerLogoMap[Providers.AIAND]).toContain("aiand"); + expect(getPlaceholder(Providers.AIAND)).toBe("aiand/openai/gpt-oss-120b"); + }); + it("maps Microsoft 365 Copilot to its chat model provider and placeholder", () => { expect(provider_map.MICROSOFT_365_COPILOT).toBe("microsoft_365_copilot"); expect(getPlaceholder(Providers.MICROSOFT_365_COPILOT)).toBe("microsoft_365_copilot/chat"); diff --git a/ui/litellm-dashboard/src/components/provider_info_helpers.tsx b/ui/litellm-dashboard/src/components/provider_info_helpers.tsx index 1ed32c160e6..6448bd936cc 100644 --- a/ui/litellm-dashboard/src/components/provider_info_helpers.tsx +++ b/ui/litellm-dashboard/src/components/provider_info_helpers.tsx @@ -1,6 +1,7 @@ import { resolveLogoSrc } from "@/lib/assetPaths"; import a2aAgentLogo from "../../public/assets/logos/a2a_agent.png"; import ai21Logo from "../../public/assets/logos/ai21.svg"; +import aiandLogo from "../../public/assets/logos/aiand.svg"; import aimlApiLogo from "../../public/assets/logos/aiml_api.svg"; import anthropicLogo from "../../public/assets/logos/anthropic.svg"; import assemblyaiSmallLogo from "../../public/assets/logos/assemblyai_small.png"; @@ -76,6 +77,7 @@ export enum Providers { A2A_Agent = "A2A Agent", AI21 = "Ai21", AI21_CHAT = "Ai21 Chat", + AIAND = "ai&", AIML = "AI/ML API", AIOHTTP_OPENAI = "Aiohttp Openai", Anthropic = "Anthropic", @@ -207,6 +209,7 @@ export const provider_map: Record = { A2A_Agent: "a2a_agent", AI21: "ai21", AI21_CHAT: "ai21_chat", + AIAND: "aiand", AIML: "aiml", AIOHTTP_OPENAI: "aiohttp_openai", Anthropic: "anthropic", @@ -344,6 +347,7 @@ export const providerLogoMap: Partial> = { [Providers.A2A_Agent]: a2aAgentLogo.src, [Providers.AI21]: ai21Logo.src, [Providers.AI21_CHAT]: ai21Logo.src, + [Providers.AIAND]: aiandLogo.src, [Providers.AIML]: aimlApiLogo.src, [Providers.AIOHTTP_OPENAI]: openaiSmallLogo.src, [Providers.Anthropic]: anthropicLogo.src, @@ -472,6 +476,7 @@ export const getProviderLogoAndName = (providerValue: string): { logo: string; d const providerPlaceholderMap: Partial> = { [Providers.AIML]: "aiml/flux-pro/v1.1", + [Providers.AIAND]: "aiand/openai/gpt-oss-120b", [Providers.Anthropic]: "claude-3-opus", [Providers.Azure]: "my-deployment", [Providers.Azure_AI_Studio]: "azure_ai/command-r-plus", diff --git a/ui/litellm-dashboard/src/lib/logoTreatments.ts b/ui/litellm-dashboard/src/lib/logoTreatments.ts index fc19c5c364d..4cd6f65a613 100644 --- a/ui/litellm-dashboard/src/lib/logoTreatments.ts +++ b/ui/litellm-dashboard/src/lib/logoTreatments.ts @@ -32,6 +32,7 @@ const TREATMENT_BY_ASSET: Readonly> = { "v0.svg": "invert", "vercel.svg": "invert", "watsonx.svg": "invert", + "aiand.svg": "plate", "aiml_api.svg": "plate", "akto.svg": "plate", "aws.svg": "plate", From ccdc8c7d10f8b042f576b1adfb670a02011cddcd Mon Sep 17 00:00:00 2001 From: mrinal-berri Date: Fri, 30 Oct 2026 10:59:28 -0700 Subject: [PATCH 17/18] docs(daily-activity): describe guardrail spend fields and reporting key IDs in OpenAPI (#49236) * docs(daily-activity): describe guardrail spend and reporting key IDs Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * docs(daily-activity): describe reporting key IDs on /user/daily/activity Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> --------- Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> --- litellm/proxy/_lazy_openapi_snapshot.json | 17 +++++++-- .../daily_activity_scopes.py | 6 ++- .../internal_user_endpoints.py | 3 +- .../common_daily_activity.py | 27 +++++++++----- ui/litellm-dashboard/src/lib/http/schema.d.ts | 37 +++++++++++++------ 5 files changed, 64 insertions(+), 26 deletions(-) diff --git a/litellm/proxy/_lazy_openapi_snapshot.json b/litellm/proxy/_lazy_openapi_snapshot.json index c76b9be7c33..dc8b9415020 100644 --- a/litellm/proxy/_lazy_openapi_snapshot.json +++ b/litellm/proxy/_lazy_openapi_snapshot.json @@ -3491,6 +3491,7 @@ "additionalProperties": { "$ref": "#/components/schemas/GuardrailMetric" }, + "description": "Spend by guardrail ID, already included in key and model spend. Populated on user, team and organization aggregated activity only, without a model filter", "title": "Guardrails", "type": "object" }, @@ -3812,10 +3813,12 @@ "GuardrailKeyMetric": { "properties": { "metadata": { - "$ref": "#/components/schemas/KeyMetadata" + "$ref": "#/components/schemas/KeyMetadata", + "description": "API key alias and existence information" }, "metrics": { - "$ref": "#/components/schemas/GuardrailSpendMetrics" + "$ref": "#/components/schemas/GuardrailSpendMetrics", + "description": "Guardrail spend and unpriced units for this API key" } }, "required": [ @@ -3828,6 +3831,7 @@ "GuardrailMetadata": { "properties": { "guardrail_name": { + "description": "Guardrail display name, falling back to the guardrail ID", "title": "Guardrail Name", "type": "string" } @@ -3844,14 +3848,17 @@ "additionalProperties": { "$ref": "#/components/schemas/GuardrailKeyMetric" }, + "description": "Per-key guardrail spend and unpriced units, limited to keys in the day's API key breakdown", "title": "Api Key Breakdown", "type": "object" }, "metadata": { - "$ref": "#/components/schemas/GuardrailMetadata" + "$ref": "#/components/schemas/GuardrailMetadata", + "description": "Guardrail name information" }, "metrics": { - "$ref": "#/components/schemas/GuardrailSpendMetrics" + "$ref": "#/components/schemas/GuardrailSpendMetrics", + "description": "Guardrail spend and unpriced units across all matching API keys" } }, "required": [ @@ -3865,11 +3872,13 @@ "properties": { "spend": { "default": 0.0, + "description": "Guardrail spend in USD, already included in key and model spend", "title": "Spend", "type": "number" }, "untracked_units": { "default": 0, + "description": "Guardrail units with no known price, excluded from spend", "title": "Untracked Units", "type": "integer" } diff --git a/litellm/proxy/management_endpoints/daily_activity_scopes.py b/litellm/proxy/management_endpoints/daily_activity_scopes.py index ca1a96f9d84..7ff73ac6e2e 100644 --- a/litellm/proxy/management_endpoints/daily_activity_scopes.py +++ b/litellm/proxy/management_endpoints/daily_activity_scopes.py @@ -216,7 +216,11 @@ def _user_query( start_date: str | None = Query(default=None, description="Start date in YYYY-MM-DD format"), end_date: str | None = Query(default=None, description="End date in YYYY-MM-DD format"), model: str | None = Query(default=None, description="Filter by specific model"), - api_key: str | None = Query(default=None, description="Filter by specific API key"), + api_key: str | None = Query( + default=None, + description="Filter by reporting API key ID. Missing keys are reported as unattributed. " + "For a literal key unattributed or a key starting with raw:, add an extra raw: prefix", + ), user_id: str | None = Query( default=None, description="Filter by specific user ID. Admins can filter by any user or omit for global view. " diff --git a/litellm/proxy/management_endpoints/internal_user_endpoints.py b/litellm/proxy/management_endpoints/internal_user_endpoints.py index 6950cf08a25..54a2d4062d2 100644 --- a/litellm/proxy/management_endpoints/internal_user_endpoints.py +++ b/litellm/proxy/management_endpoints/internal_user_endpoints.py @@ -2922,7 +2922,8 @@ async def get_user_daily_activity( ), api_key: str | None = fastapi.Query( default=None, - description="Filter by specific API key", + description="Filter by reporting API key ID. Missing keys are reported as unattributed. " + "For a literal key unattributed or a key starting with raw:, add an extra raw: prefix", ), user_id: str | None = fastapi.Query( default=None, diff --git a/litellm/types/proxy/management_endpoints/common_daily_activity.py b/litellm/types/proxy/management_endpoints/common_daily_activity.py index 3f1a7980b27..37f59711050 100644 --- a/litellm/types/proxy/management_endpoints/common_daily_activity.py +++ b/litellm/types/proxy/management_endpoints/common_daily_activity.py @@ -68,29 +68,34 @@ class MetricWithMetadata(MetricBase): class GuardrailSpendMetrics(LiteLLMBaseModel): model_config = ConfigDict(frozen=True) - spend: float = 0.0 - untracked_units: int = 0 + spend: float = Field(default=0.0, description="Guardrail spend in USD, already included in key and model spend") + untracked_units: int = Field(default=0, description="Guardrail units with no known price, excluded from spend") class GuardrailKeyMetric(LiteLLMBaseModel): model_config = ConfigDict(frozen=True) - metrics: GuardrailSpendMetrics - metadata: KeyMetadata + metrics: GuardrailSpendMetrics = Field(description="Guardrail spend and unpriced units for this API key") + metadata: KeyMetadata = Field(description="API key alias and existence information") class GuardrailMetadata(LiteLLMBaseModel): model_config = ConfigDict(frozen=True) - guardrail_name: str + guardrail_name: str = Field(description="Guardrail display name, falling back to the guardrail ID") class GuardrailMetric(LiteLLMBaseModel): model_config = ConfigDict(frozen=True) - metrics: GuardrailSpendMetrics - metadata: GuardrailMetadata - api_key_breakdown: Mapping[str, GuardrailKeyMetric] = Field(default_factory=dict) + metrics: GuardrailSpendMetrics = Field( + description="Guardrail spend and unpriced units across all matching API keys" + ) + metadata: GuardrailMetadata = Field(description="Guardrail name information") + api_key_breakdown: Mapping[str, GuardrailKeyMetric] = Field( + default_factory=dict, + description="Per-key guardrail spend and unpriced units, limited to keys in the day's API key breakdown", + ) class BreakdownMetrics(LiteLLMBaseModel): @@ -103,7 +108,11 @@ class BreakdownMetrics(LiteLLMBaseModel): endpoints: dict[str, MetricWithMetadata] = Field(default_factory=dict) # endpoint -> {metrics, metadata} api_keys: dict[str, KeyMetricWithMetadata] = Field(default_factory=dict) # api_key -> {metrics, metadata} entities: dict[str, MetricWithMetadata] = Field(default_factory=dict) # entity -> {metrics, metadata} - guardrails: Mapping[str, GuardrailMetric] = Field(default_factory=dict) # guardrail_id -> spend + guardrails: Mapping[str, GuardrailMetric] = Field( + default_factory=dict, + description="Spend by guardrail ID, already included in key and model spend. " + "Populated on user, team and organization aggregated activity only, without a model filter", + ) class DailySpendData(LiteLLMBaseModel): diff --git a/ui/litellm-dashboard/src/lib/http/schema.d.ts b/ui/litellm-dashboard/src/lib/http/schema.d.ts index 960616e2631..aafa732276b 100644 --- a/ui/litellm-dashboard/src/lib/http/schema.d.ts +++ b/ui/litellm-dashboard/src/lib/http/schema.d.ts @@ -27870,7 +27870,10 @@ export interface components { entities?: { [key: string]: components["schemas"]["MetricWithMetadata"]; }; - /** Guardrails */ + /** + * Guardrails + * @description Spend by guardrail ID, already included in key and model spend. Populated on user, team and organization aggregated activity only, without a model filter + */ guardrails?: { [key: string]: components["schemas"]["GuardrailMetric"]; }; @@ -33448,32 +33451,44 @@ export interface components { }; /** GuardrailKeyMetric */ GuardrailKeyMetric: { + /** @description API key alias and existence information */ metadata: components["schemas"]["KeyMetadata"]; + /** @description Guardrail spend and unpriced units for this API key */ metrics: components["schemas"]["GuardrailSpendMetrics"]; }; /** GuardrailMetadata */ GuardrailMetadata: { - /** Guardrail Name */ + /** + * Guardrail Name + * @description Guardrail display name, falling back to the guardrail ID + */ guardrail_name: string; }; /** GuardrailMetric */ GuardrailMetric: { - /** Api Key Breakdown */ + /** + * Api Key Breakdown + * @description Per-key guardrail spend and unpriced units, limited to keys in the day's API key breakdown + */ api_key_breakdown?: { [key: string]: components["schemas"]["GuardrailKeyMetric"]; }; + /** @description Guardrail name information */ metadata: components["schemas"]["GuardrailMetadata"]; + /** @description Guardrail spend and unpriced units across all matching API keys */ metrics: components["schemas"]["GuardrailSpendMetrics"]; }; /** GuardrailSpendMetrics */ GuardrailSpendMetrics: { /** * Spend + * @description Guardrail spend in USD, already included in key and model spend * @default 0 */ spend: number; /** * Untracked Units + * @description Guardrail units with no known price, excluded from spend * @default 0 */ untracked_units: number; @@ -76633,7 +76648,7 @@ export interface operations { end_date?: string | null; /** @description Filter by specific model */ model?: string | null; - /** @description Filter by specific API key */ + /** @description Filter by reporting API key ID. Missing keys are reported as unattributed. For a literal key unattributed or a key starting with raw:, add an extra raw: prefix */ api_key?: string | null; /** @description Filter by specific user ID. Admins can filter by any user or omit for global view. Non-admins must provide their own user_id. */ user_id?: string | null; @@ -76682,7 +76697,7 @@ export interface operations { end_date?: string | null; /** @description Filter by specific model */ model?: string | null; - /** @description Filter by specific API key */ + /** @description Filter by reporting API key ID. Missing keys are reported as unattributed. For a literal key unattributed or a key starting with raw:, add an extra raw: prefix */ api_key?: string | null; /** @description Filter by specific user ID. Admins can filter by any user or omit for global view. Non-admins must provide their own user_id. */ user_id?: string | null; @@ -76727,7 +76742,7 @@ export interface operations { end_date?: string | null; /** @description Filter by specific model */ model?: string | null; - /** @description Filter by specific API key */ + /** @description Filter by reporting API key ID. Missing keys are reported as unattributed. For a literal key unattributed or a key starting with raw:, add an extra raw: prefix */ api_key?: string | null; /** @description Filter by specific user ID. Admins can filter by any user or omit for global view. Non-admins must provide their own user_id. */ user_id?: string | null; @@ -76773,7 +76788,7 @@ export interface operations { end_date?: string | null; /** @description Filter by specific model */ model?: string | null; - /** @description Filter by specific API key */ + /** @description Filter by reporting API key ID. Missing keys are reported as unattributed. For a literal key unattributed or a key starting with raw:, add an extra raw: prefix */ api_key?: string | null; /** @description Filter by specific user ID. Admins can filter by any user or omit for global view. Non-admins must provide their own user_id. */ user_id?: string | null; @@ -76820,7 +76835,7 @@ export interface operations { end_date?: string | null; /** @description Filter by specific model */ model?: string | null; - /** @description Filter by specific API key */ + /** @description Filter by reporting API key ID. Missing keys are reported as unattributed. For a literal key unattributed or a key starting with raw:, add an extra raw: prefix */ api_key?: string | null; /** @description Filter by specific user ID. Admins can filter by any user or omit for global view. Non-admins must provide their own user_id. */ user_id?: string | null; @@ -76866,7 +76881,7 @@ export interface operations { end_date?: string | null; /** @description Filter by specific model */ model?: string | null; - /** @description Filter by specific API key */ + /** @description Filter by reporting API key ID. Missing keys are reported as unattributed. For a literal key unattributed or a key starting with raw:, add an extra raw: prefix */ api_key?: string | null; /** @description Filter by specific user ID. Admins can filter by any user or omit for global view. Non-admins must provide their own user_id. */ user_id?: string | null; @@ -76912,7 +76927,7 @@ export interface operations { end_date?: string | null; /** @description Filter by specific model */ model?: string | null; - /** @description Filter by specific API key */ + /** @description Filter by reporting API key ID. Missing keys are reported as unattributed. For a literal key unattributed or a key starting with raw:, add an extra raw: prefix */ api_key?: string | null; /** @description Filter by specific user ID. Admins can filter by any user or omit for global view. Non-admins must provide their own user_id. */ user_id?: string | null; @@ -76958,7 +76973,7 @@ export interface operations { end_date?: string | null; /** @description Filter by specific model */ model?: string | null; - /** @description Filter by specific API key */ + /** @description Filter by reporting API key ID. Missing keys are reported as unattributed. For a literal key unattributed or a key starting with raw:, add an extra raw: prefix */ api_key?: string | null; /** @description Filter by specific user ID. Admins can filter by any user or omit for global view. Non-admins must provide their own user_id. */ user_id?: string | null; From f7f87e6193e151db09ce3db40f26dcff5333a59b Mon Sep 17 00:00:00 2001 From: Ahmet Selim Fedakar <122402435+selimfedakar@users.noreply.github.com> Date: Fri, 30 Oct 2026 21:01:00 +0300 Subject: [PATCH 18/18] fix(responses): normalize empty function_call_output to an empty tool result (#48845) * fix(responses): normalize empty function_call_output to an empty tool result * test(responses): preserve empty tool outputs through guardrails --------- Co-authored-by: Tin --- .../transformation.py | 10 ++++--- ...test_openai_responses_guardrail_handler.py | 14 +++++++--- .../test_anthropic_responses_bridge.py | 26 +++++++++++++++++ ...test_function_call_output_normalization.py | 17 ++++++++++- .../test_google_ai_studio_responses_wire.py | 28 +++++++++++++++++++ 5 files changed, 86 insertions(+), 9 deletions(-) diff --git a/litellm/responses/litellm_completion_transformation/transformation.py b/litellm/responses/litellm_completion_transformation/transformation.py index ce8841c9190..8f0a85d67de 100644 --- a/litellm/responses/litellm_completion_transformation/transformation.py +++ b/litellm/responses/litellm_completion_transformation/transformation.py @@ -398,23 +398,25 @@ def tool_output_to_chat_content( if single_media is not None: return [single_media] if isinstance(output, list): + parts: Final = cast(Sequence[object], output) # cast-ok: checked list shape, elements remain unvalidated normalized_blocks: Final[list[ToolMessageContentPart]] = [] text_acc: Final[list[str]] = [] - for part in output: + for part in parts: if not isinstance(part, dict): continue part_type = part.get("type") if part_type in ("input_text", "output_text", "text"): txt = part.get("text") - if isinstance(txt, str) and txt: + if isinstance(txt, str): text_acc.append(txt) - normalized_blocks.append({"type": "text", "text": txt}) + if txt: + normalized_blocks.append({"type": "text", "text": txt}) elif (media := decode_responses_media(part)) is not None: normalized_blocks.append(media) if any(b.get("type") != "text" for b in normalized_blocks): return normalized_blocks - if text_acc: + if any(text_acc) or len(text_acc) == len(parts): return "".join(text_acc) try: return json.dumps(output) diff --git a/tests/unit/llms/openai/responses/test_openai_responses_guardrail_handler.py b/tests/unit/llms/openai/responses/test_openai_responses_guardrail_handler.py index 673e880a1c0..d732d922375 100644 --- a/tests/unit/llms/openai/responses/test_openai_responses_guardrail_handler.py +++ b/tests/unit/llms/openai/responses/test_openai_responses_guardrail_handler.py @@ -724,16 +724,23 @@ class TestToolResultItemsReachTextsGuardrails: assert blocked.value.status_code == 400 @pytest.mark.asyncio - @pytest.mark.parametrize("empty_output", ["", None], ids=["empty_string", "null"]) - async def test_empty_tool_output_sends_no_text_and_stays_as_sent(self, empty_output: str | None) -> None: + @pytest.mark.parametrize(("call_item", "output_type"), _TOOL_TURNS, ids=["function", "custom"]) + @pytest.mark.parametrize( + "empty_output", + ["", None, [], *[[{"type": kind, "text": ""}] for kind in ("input_text", "output_text", "text")]], + ids=["empty_string", "null", "empty_list", "input_text", "output_text", "text"], + ) + async def test_empty_tool_output_sends_no_text_and_stays_as_sent( + self, call_item: dict[str, str], output_type: str, empty_output: object + ) -> None: guardrail: Final = RecordingMaskingGuardrail(guardrail_name="test") - call_item, output_type = _TOOL_TURNS[0] result: Final = await OpenAIResponsesHandler().process_input_messages( _tool_result_request(call_item, output_type, empty_output), guardrail ) assert guardrail.seen_texts == [["Weather?"]] + assert guardrail.seen_message_contents == [["Weather?", None]] assert result["input"][2] == {"type": output_type, "call_id": "call_1", "output": empty_output} @pytest.mark.asyncio @@ -1103,7 +1110,6 @@ class TestToolResultItemsReachTextsGuardrails: ["visible"], id="non_dict_part", ), - pytest.param([], ["[]"], id="empty"), pytest.param({"result": "SECRET forecast"}, ['{"result": "SECRET forecast"}'], id="dict"), pytest.param( [{"type": "input_image", "file_id": "file-map"}], diff --git a/tests/unit/responses/litellm_completion_transformation/test_anthropic_responses_bridge.py b/tests/unit/responses/litellm_completion_transformation/test_anthropic_responses_bridge.py index ef198b487b9..0341511496c 100644 --- a/tests/unit/responses/litellm_completion_transformation/test_anthropic_responses_bridge.py +++ b/tests/unit/responses/litellm_completion_transformation/test_anthropic_responses_bridge.py @@ -23,6 +23,7 @@ from litellm.responses.litellm_completion_transformation.transformation import ( ) from litellm.types.llms.openai import ResponsesAPIStreamEvents from litellm.types.utils import Message, ModelResponse +from tests.unit.proxy.conftest import httpx_transport def test_response_api_handler_merges_metadata_and_service_tier_without_error(): @@ -593,6 +594,31 @@ def _anthropic_message(*content: Mapping[str, object]) -> httpx.Response: ) +@pytest.mark.usefixtures(httpx_transport.__name__) +@pytest.mark.parametrize("output", ["", [], [{"type": "input_text", "text": ""}]]) +def test_empty_responses_tool_output_reaches_anthropic_without_json_text( + respx_mock: respx.MockRouter, output: str | list[dict[str, str]] +) -> None: + route: Final = respx_mock.post(_ANTHROPIC_MESSAGES_URL).mock( + return_value=_anthropic_message({"type": "text", "text": "ok"}) + ) + response: Final = litellm.responses( + model=_MODEL, + api_key="mock_api_key", + input=[ + {"role": "user", "content": "Check the weather"}, + {"type": "function_call", "call_id": "call_empty", "name": "get_weather", "arguments": "{}"}, + {"type": "function_call_output", "call_id": "call_empty", "output": output}, + ], + ) + sent: Final = json.loads(route.calls.last.request.content) + assert response.output_text == "ok" + assert sent["messages"][-1] == { + "role": "user", + "content": [{"type": "tool_result", "tool_use_id": "call_empty", "content": ""}], + } + + def _replay(output: Sequence[BaseModel]) -> list[object]: return [*_FIRST_INPUT, *(item.model_dump(exclude_none=True) for item in output), *_NEXT_INPUT] diff --git a/tests/unit/responses/litellm_completion_transformation/test_function_call_output_normalization.py b/tests/unit/responses/litellm_completion_transformation/test_function_call_output_normalization.py index 65800e42124..fce06598d29 100644 --- a/tests/unit/responses/litellm_completion_transformation/test_function_call_output_normalization.py +++ b/tests/unit/responses/litellm_completion_transformation/test_function_call_output_normalization.py @@ -50,9 +50,10 @@ def test_image_detail_mime_and_adjacent_text_survive_normalization(kind: str, de else "data:application/octet-stream;base64,AAAA", **({} if nested_url else metadata), } - parts: Final = [{"type": "input_text", "text": "Screenshot follows"}, image] + parts: Final = [{"type": "input_text", "text": ""}, {"type": "input_text", "text": "Screenshot follows"}, image] original: Final = deepcopy(parts) assert _normalize_media_content(kind, parts) == [ + *([{"type": "text", "text": ""}] if kind == "direct" else []), {"type": "text", "text": "Screenshot follows"}, {"type": "image_url", "image_url": {"url": "data:application/octet-stream;base64,AAAA", **metadata}}, ] @@ -97,6 +98,12 @@ def test_media_normalization_preserves_formats_references_and_metadata(kind: str ("output", "expected"), [ (None, ""), + ("", ""), + ([], ""), + ([{"type": "input_text", "text": ""}], ""), + ([{"type": kind, "text": ""} for kind in ("input_text", "output_text", "text")], ""), + ("[]", "[]"), + ([{"type": "input_text", "text": " \n"}], " \n"), ('{"ok":true}', '{"ok":true}'), ([{"type": "input_text", "text": "hello"}, {"type": "input_text", "text": " world"}], "hello world"), ({"type": "file", "status": "ready"}, '{"type": "file", "status": "ready"}'), @@ -106,6 +113,14 @@ def test_media_normalization_preserves_formats_references_and_metadata(kind: str ), ({"type": "input_image", "format": 17}, '{"type": "input_image", "format": 17}'), ([{"type": "unknown", "value": 17}], '[{"type": "unknown", "value": 17}]'), + ([{"type": "input_text", "text": 17}], '[{"type": "input_text", "text": 17}]'), + ([{"type": "input_text"}], '[{"type": "input_text"}]'), + ([{"type": "input_text", "text": None}], '[{"type": "input_text", "text": null}]'), + ( + [{"type": "input_text", "text": ""}, {"type": "unknown", "value": 17}], + '[{"type": "input_text", "text": ""}, {"type": "unknown", "value": 17}]', + ), + ([{"type": "input_text", "text": ""}, 17], '[{"type": "input_text", "text": ""}, 17]'), ( [{"type": "input_text", "text": "keep"}, {"type": "input_image", "file_id": "file-only"}], "keep", diff --git a/tests/unit/responses/litellm_completion_transformation/test_google_ai_studio_responses_wire.py b/tests/unit/responses/litellm_completion_transformation/test_google_ai_studio_responses_wire.py index f63088a606f..c7acc70ba95 100644 --- a/tests/unit/responses/litellm_completion_transformation/test_google_ai_studio_responses_wire.py +++ b/tests/unit/responses/litellm_completion_transformation/test_google_ai_studio_responses_wire.py @@ -84,6 +84,34 @@ async def test_web_search_preview_is_sent_as_google_search() -> None: assert request_body["tools"] == [{"googleSearch": {}}] +@pytest.mark.asyncio +@pytest.mark.parametrize("output", ["", [], [{"type": "input_text", "text": ""}]]) +async def test_empty_responses_tool_output_reaches_gemini_without_json_text( + output: str | list[dict[str, str]], +) -> None: + with respx.mock() as mock_router: + route: Final = mock_router.post(url__regex=_GEMINI_URL).mock( + return_value=httpx.Response(status_code=200, json=_function_call_response("signature")) + ) + response: Final = await litellm.aresponses( + model="gemini/gemini-2.5-flash", + api_key="test-key", + input=[ + {"role": "user", "content": "Check the weather"}, + {"type": "function_call", "call_id": "call_empty", "name": "get_weather", "arguments": "{}"}, + {"type": "function_call_output", "call_id": "call_empty", "output": output}, + ], + ) + request_body: Final = _JSON_OBJECT.validate_json(route.calls.last.request.content) + + assert isinstance(response, ResponsesAPIResponse) + assert request_body["contents"] == [ + {"role": "user", "parts": [{"text": "Check the weather"}]}, + {"role": "model", "parts": [{"function_call": {"name": "get_weather", "args": {}}}]}, + {"role": "user", "parts": [{"function_response": {"name": "get_weather", "response": {"content": ""}}}]}, + ] + + @pytest.mark.asyncio @pytest.mark.parametrize( ("assistant_content", "assistant_parts"),