fix(responses): drop params compact rejects and send Azure compaction to the v1 route (#48876)

* fix(responses): drop metadata from compact requests to OpenAI and Azure Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(responses): drop user and drop_params-rejected fields on compact and route Azure compaction to v1 * fix(responses): keep stream on compact and keep non-dated Azure api-versions on the v1 compact route * fix(responses): drop compact params only for providers verified to reject them --------- Co-authored-by: mateo <mateo@berri.ai> Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com>
2026-10-29 06:33:28 +00:00 · 2026-10-28 19:46:36 -07:00 · 2026-10-28 19:46:36 -07:00 · 83ae1bd63e
commit 83ae1bd63e
parent 4fc2bba1f2
7 changed files with 241 additions and 27 deletions
--- a/litellm/llms/azure/common_utils.py
+++ b/litellm/llms/azure/common_utils.py
@ -621,7 +621,7 @@ class BaseAzureLLM(BaseOpenAILLM):

        # For Azure v1 API, use standard OpenAI client instead of AzureOpenAI
        # See: https://learn.microsoft.com/en-us/azure/ai-services/openai/reference#api-specs
-        if self._is_azure_v1_api_version(api_version):
+        if self.is_azure_v1_api_version(api_version):
            # Extract only params that OpenAI client accepts
            # Always use /openai/v1/ regardless of whether user passed "v1", "latest", or "preview"
            # The OpenAI client accepts a callable for `api_key` and re-invokes it
@ -922,7 +922,7 @@ class BaseAzureLLM(BaseOpenAILLM):
        else:
            new_url = api_base

-        if BaseAzureLLM._is_azure_v1_api_version(api_version):
+        if BaseAzureLLM.is_azure_v1_api_version(api_version):
            # ensure the request go to /openai/v1 and not just /openai
            if "/openai/v1" not in new_url:
                parsed_url: Final = httpx.URL(new_url)
@ -952,7 +952,7 @@ class BaseAzureLLM(BaseOpenAILLM):

        Returns None when ``api_version`` is a dated one, which still uses the deployment route.
        """
-        if not BaseAzureLLM._is_azure_v1_api_version(api_version):
+        if not BaseAzureLLM.is_azure_v1_api_version(api_version):
            return None

        base_url: Final = httpx.URL(api_base)
@ -970,7 +970,7 @@ class BaseAzureLLM(BaseOpenAILLM):
        )

    @staticmethod
-    def _is_azure_v1_api_version(api_version: str | None) -> bool:
+    def is_azure_v1_api_version(api_version: str | None) -> bool:
        if api_version is None:
            return False
        return api_version in {"preview", "latest", "v1"}
--- a/litellm/llms/azure/responses/transformation.py
+++ b/litellm/llms/azure/responses/transformation.py
@ -54,6 +54,16 @@ def _resolve_responses_api_version(api_base: str | None, litellm_params: Mapping
    return api_version


+def _v1_responses_url(api_base: str) -> httpx.URL:
+    parsed_url: Final = httpx.URL(api_base)
+    path: Final = parsed_url.path.rstrip("/")
+    root: Final = next(
+        (path.removesuffix(route) for route in ("/openai/v1/responses", "/openai/responses") if path.endswith(route)),
+        path,
+    )
+    return parsed_url.copy_with(path=f"{root}/openai/v1/responses", query=None)
+
+
 def _with_responses_api_version(api_base: str | None, litellm_params: Mapping[str, object]) -> Mapping[str, object]:
    return MappingProxyType({**litellm_params, "api_version": _resolve_responses_api_version(api_base, litellm_params)})

@ -247,20 +257,20 @@ class AzureOpenAIResponsesAPIConfig(OpenAIResponsesAPIConfig):
        if api_base is None:
            raise ValueError("api_base is required for Azure WebSocket")

-        parsed_url: Final = httpx.URL(api_base)
-        path = parsed_url.path.rstrip("/")
-        # Strip existing /openai/responses path if the api_base already contains it
-        for suffix in ("/openai/v1/responses", "/openai/responses"):
-            if path.endswith(suffix):
-                path = path[: -len(suffix)]
-                break
-        scheme: Final = "wss" if parsed_url.scheme == "https" else "ws"
-        return str(parsed_url.copy_with(scheme=scheme, path=f"{path}/openai/v1/responses", query=None))
+        v1_url: Final = _v1_responses_url(api_base)
+        return str(v1_url.copy_with(scheme="wss" if v1_url.scheme == "https" else "ws"))

    def model_in_websocket_url(self) -> bool:
        # Azure sends the model in the response.create body, not the URL
        return False

+    def compact_url(self, api_base: str) -> str:
+        params: Final = httpx.URL(api_base).params
+        v1_params: Final = (
+            params if BaseAzureLLM.is_azure_v1_api_version(params.get("api-version")) else params.remove("api-version")
+        )
+        return super().compact_url(str(_v1_responses_url(api_base).copy_with(params=v1_params)))
+
    #########################################################
    ########## DELETE RESPONSE API TRANSFORMATION ##############
    #########################################################
--- a/litellm/llms/openai/responses/transformation.py
+++ b/litellm/llms/openai/responses/transformation.py
@ -66,6 +66,29 @@ _PROVIDERS_VALIDATING_TOOL_CALL_ITEM_IDS: Final = frozenset({LlmProviders.AZURE,
 _PROVIDERS_REPLAYING_ONLY_THEIR_OWN_REASONING: Final = _PROVIDERS_VALIDATING_TOOL_CALL_ITEM_IDS
 _ROLES_READING_PROMPT_CACHE_BREAKPOINTS: Final = frozenset({"system", "developer", "user"})
 _PART_TYPES_READING_PROMPT_CACHE_BREAKPOINTS: Final = frozenset({"input_text", "input_image", "input_file"})
+_COMPACT_TRACKING_PARAMS: Final = frozenset({"metadata", "user"})
+_COMPACT_REJECTED_PARAMS: Final = frozenset(
+    {
+        "background",
+        "context_management",
+        "max_output_tokens",
+        "max_tool_calls",
+        "moderation",
+        "partial_images",
+        "prompt",
+        "safety_identifier",
+        "store",
+        "stream_options",
+        "temperature",
+        "tool_choice",
+        "top_logprobs",
+        "top_p",
+        "truncation",
+    }
+)
+_PROVIDERS_WITH_STRICT_COMPACT_PARAMS: Final = frozenset(
+    {LlmProviders.OPENAI, LlmProviders.AZURE, LlmProviders.AZURE_AI}
+)
 _INPUT_ITEM: Final[TypeAdapter[Mapping[str, object]]] = TypeAdapter(Mapping[str, object])
 _CONTENT_PARTS: Final[TypeAdapter[Sequence[Mapping[str, object]]]] = TypeAdapter(Sequence[Mapping[str, object]])

@ -1182,6 +1205,17 @@ class OpenAIResponsesAPIConfig(BaseResponsesAPIConfig):
    #########################################################
    ########## COMPACT RESPONSE API TRANSFORMATION ##########
    #########################################################
+    def compact_url(self, api_base: str) -> str:
+        parsed_url: Final = httpx.URL(api_base)
+        return str(parsed_url.copy_with(path=parsed_url.path.rstrip("/") + "/compact"))
+
+    def compact_dropped_params(self, litellm_params: GenericLiteLLMParams) -> frozenset[str]:
+        if self.custom_llm_provider not in _PROVIDERS_WITH_STRICT_COMPACT_PARAMS:
+            return frozenset()
+        if _drops_params(litellm_params):
+            return _COMPACT_TRACKING_PARAMS | _COMPACT_REJECTED_PARAMS
+        return _COMPACT_TRACKING_PARAMS
+
    def transform_compact_response_api_request(
        self,
        model: str,
@ -1197,10 +1231,8 @@ class OpenAIResponsesAPIConfig(BaseResponsesAPIConfig):
        OpenAI API expects the following request
        - POST /v1/responses/compact
        """
-        # Preserve query params (e.g., api-version) while appending /compact.
-        parsed_url: Final = httpx.URL(api_base)
-        compact_path: Final = parsed_url.path.rstrip("/") + "/compact"
-        url: Final = str(parsed_url.copy_with(path=compact_path))
+        url: Final = self.compact_url(api_base)
+        dropped_params: Final = self.compact_dropped_params(litellm_params)

        request_params: Final = cast(  # cast-ok: the base signature leaves the params bag untyped
            "Mapping[str, object]", response_api_optional_request_params
@ -1214,14 +1246,11 @@ class OpenAIResponsesAPIConfig(BaseResponsesAPIConfig):
        )
        if sanitized_tools is not None:
            response_api_optional_request_params["tools"] = sanitized_tools
-        data: Final = cast(  # cast-ok: typed dict spread is inferred as Unknown by basedpyright
-            dict[str, object],
-            {
-                "model": model,
-                "input": replay_safe_input,
-                **response_api_optional_request_params,
-            },
-        )
+        data: Final = {
+            "model": model,
+            "input": replay_safe_input,
+            **{key: value for key, value in request_params.items() if key not in dropped_params},
+        }

        return url, data

--- a/tests/unit/llms/azure/response/test_azure_transformation.py
+++ b/tests/unit/llms/azure/response/test_azure_transformation.py
@ -165,6 +165,44 @@ def test_explicit_v1_api_base_keeps_responses_api_version_selection(


@pytest.mark.serial
+@pytest.mark.parametrize(
+    ("api_base", "api_version", "expected_url"),
+    [
+        (
+            "https://x.openai.azure.com",
+            "2025-04-01-preview",
+            "https://x.openai.azure.com/openai/v1/responses/compact",
+        ),
+        (
+            "https://x.openai.azure.com",
+            "preview",
+            "https://x.openai.azure.com/openai/v1/responses/compact?api-version=preview",
+        ),
+        (
+            "https://gw.example.com/foundry/openai/openai/v1?tenant=example",
+            "v1",
+            "https://gw.example.com/foundry/openai/openai/v1/responses/compact?tenant=example&api-version=v1",
+        ),
+    ],
+)
+def test_compact_reaches_the_v1_route_and_drops_only_a_dated_api_version(
+    api_base: str, api_version: str, expected_url: str
+) -> None:
+    config: Final = AzureOpenAIResponsesAPIConfig()
+    responses_url: Final = config.get_complete_url(api_base=api_base, litellm_params={"api_version": api_version})
+
+    url, _ = config.transform_compact_response_api_request(
+        model="gpt-5.4-mini",
+        input="hi",
+        response_api_optional_request_params={},
+        api_base=responses_url,
+        litellm_params=GenericLiteLLMParams(),
+        headers={},
+    )
+
+    assert url == expected_url
+
+
 def test_response_id_path_requests_encode_response_id():
    config = AzureOpenAIResponsesAPIConfig()
    api_base = (
--- a/tests/unit/llms/azure/test_azure_common_utils.py
+++ b/tests/unit/llms/azure/test_azure_common_utils.py
@ -1658,9 +1658,9 @@ def test_get_azure_ad_token_fallback_to_default_azure_credential(
 )
 def test_is_azure_v1_api_version(api_version, expected):
    """
-    Test that _is_azure_v1_api_version correctly identifies v1 API versions.
+    Test that is_azure_v1_api_version correctly identifies v1 API versions.
    """
-    result = BaseAzureLLM._is_azure_v1_api_version(api_version=api_version)
+    result = BaseAzureLLM.is_azure_v1_api_version(api_version=api_version)
    assert result == expected


--- a/tests/unit/llms/litellm_proxy/responses/test_transformation.py
+++ b/tests/unit/llms/litellm_proxy/responses/test_transformation.py
@ -7,6 +7,7 @@ import respx

 import litellm
 from litellm.llms.litellm_proxy.responses.transformation import LiteLLMProxyResponsesAPIConfig
+from litellm.types.router import GenericLiteLLMParams
 from litellm.types.utils import LlmProviders
 from litellm.utils import ProviderConfigManager
 from tests.unit.proxy.conftest import httpx_transport
@ -73,3 +74,23 @@ def test_get_complete_url_requires_api_base(monkeypatch: pytest.MonkeyPatch) ->
    monkeypatch.delenv("LITELLM_PROXY_API_BASE", raising=False)
    with pytest.raises(ValueError, match="api_base not set"):
        LiteLLMProxyResponsesAPIConfig().get_complete_url(api_base=None, litellm_params={})
+
+
+@pytest.mark.parametrize("drop_params", [False, True])
+def test_compact_request_forwards_metadata_and_user_to_the_downstream_proxy(drop_params: bool) -> None:
+    url, data = LiteLLMProxyResponsesAPIConfig().transform_compact_response_api_request(
+        model="gpt-5.5",
+        input="hi",
+        response_api_optional_request_params={"metadata": {"ticket": "qa-1"}, "user": "end-user-1", "temperature": 0.2},
+        api_base="https://my-proxy.example.com/responses",
+        litellm_params=GenericLiteLLMParams(drop_params=drop_params),
+        headers={},
+    )
+
+    assert url == "https://my-proxy.example.com/responses/compact"
+    assert {key: data[key] for key in data if key != "input"} == {
+        "model": "gpt-5.5",
+        "metadata": {"ticket": "qa-1"},
+        "user": "end-user-1",
+        "temperature": 0.2,
+    }
--- a/tests/unit/llms/openai/responses/test_openai_responses_transformation.py
+++ b/tests/unit/llms/openai/responses/test_openai_responses_transformation.py
@ -12,8 +12,11 @@ from pydantic import JsonValue, TypeAdapter
 import litellm
 from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
 from litellm.llms.azure.responses.transformation import AzureOpenAIResponsesAPIConfig
+from litellm.llms.azure_ai.responses.transformation import AzureAIResponsesAPIConfig
 from litellm.llms.base_llm.responses.transformation import BaseResponsesAPIConfig
+from litellm.llms.hosted_vllm.responses.transformation import HostedVLLMResponsesAPIConfig
 from litellm.llms.openai.responses.transformation import OpenAIResponsesAPIConfig
+from litellm.llms.xai.responses.transformation import XAIResponsesAPIConfig
 from litellm.responses.litellm_completion_transformation.transformation import LiteLLMCompletionResponsesConfig
 from litellm.types.llms.openai import (
    ImageGenerationPartialImageEvent,
@ -4277,3 +4280,116 @@ def test_dict_responses_input_filters_unset_reasoning_fields() -> None:
    function_call_item: Final = validated_input[2]
    assert function_call_item["type"] == "function_call"
    assert function_call_item["status"] == "completed"
+
+
+_COMPACT_CONFIGS: Final = (
+    pytest.param(OpenAIResponsesAPIConfig(), "https://api.openai.com/v1/responses", id="openai"),
+    pytest.param(
+        AzureOpenAIResponsesAPIConfig(),
+        "https://x.openai.azure.com/openai/responses?api-version=2025-04-01-preview",
+        id="azure",
+    ),
+    pytest.param(AzureAIResponsesAPIConfig(), "https://x.services.ai.azure.com/openai/v1/responses", id="azure_ai"),
+)
+
+
+@pytest.mark.parametrize(("config", "api_base"), _COMPACT_CONFIGS)
+def test_compact_request_omits_metadata_and_user_and_keeps_the_rest_without_drop_params(
+    config: OpenAIResponsesAPIConfig, api_base: str
+) -> None:
+    _, data = config.transform_compact_response_api_request(
+        model="gpt-5.4-mini",
+        input="hi",
+        response_api_optional_request_params={
+            "metadata": {"ticket": "qa-1"},
+            "user": "end-user-1",
+            "instructions": "be brief",
+            "previous_response_id": "resp_1",
+            "reasoning": {"effort": "low"},
+            "temperature": 0.2,
+        },
+        api_base=api_base,
+        litellm_params=GenericLiteLLMParams(drop_params=False),
+        headers={},
+    )
+
+    assert {key: data[key] for key in data if key != "input"} == {
+        "model": "gpt-5.4-mini",
+        "instructions": "be brief",
+        "previous_response_id": "resp_1",
+        "reasoning": {"effort": "low"},
+        "temperature": 0.2,
+    }
+
+
+@pytest.mark.parametrize(("config", "api_base"), _COMPACT_CONFIGS)
+def test_compact_request_drops_params_compact_rejects_under_drop_params(
+    config: OpenAIResponsesAPIConfig, api_base: str
+) -> None:
+    _, data = config.transform_compact_response_api_request(
+        model="gpt-5.4-mini",
+        input="hi",
+        response_api_optional_request_params={
+            "metadata": {"ticket": "qa-1"},
+            "user": "end-user-1",
+            "temperature": 0.2,
+            "store": False,
+            "truncation": "auto",
+            "max_output_tokens": 100,
+            "tool_choice": "auto",
+            "reasoning": {"effort": "low"},
+            "include": ["reasoning.encrypted_content"],
+            "text": {"format": {"type": "text"}},
+        },
+        api_base=api_base,
+        litellm_params=GenericLiteLLMParams(drop_params=True),
+        headers={},
+    )
+
+    assert {key: data[key] for key in data if key != "input"} == {
+        "model": "gpt-5.4-mini",
+        "reasoning": {"effort": "low"},
+        "include": ["reasoning.encrypted_content"],
+        "text": {"format": {"type": "text"}},
+    }
+
+
+@pytest.mark.parametrize(
+    ("config", "api_base"),
+    (
+        pytest.param(XAIResponsesAPIConfig(), "https://api.x.ai/v1/responses", id="xai"),
+        pytest.param(HostedVLLMResponsesAPIConfig(), "http://vllm.example.com/v1/responses", id="hosted_vllm"),
+    ),
+)
+def test_compact_request_forwards_every_param_for_providers_not_known_to_reject_them(
+    config: OpenAIResponsesAPIConfig, api_base: str
+) -> None:
+    _, data = config.transform_compact_response_api_request(
+        model="some-model",
+        input="hi",
+        response_api_optional_request_params={"metadata": {"ticket": "qa-1"}, "user": "end-user-1", "temperature": 0.2},
+        api_base=api_base,
+        litellm_params=GenericLiteLLMParams(drop_params=True),
+        headers={},
+    )
+
+    assert {key: data[key] for key in data if key != "input"} == {
+        "model": "some-model",
+        "metadata": {"ticket": "qa-1"},
+        "user": "end-user-1",
+        "temperature": 0.2,
+    }
+
+
+@pytest.mark.parametrize(("config", "api_base"), _COMPACT_CONFIGS)
+def test_compact_request_keeps_stream_under_drop_params(config: OpenAIResponsesAPIConfig, api_base: str) -> None:
+    _, data = config.transform_compact_response_api_request(
+        model="gpt-5.4-mini",
+        input="hi",
+        response_api_optional_request_params={"stream": True, "stream_options": {"include_obfuscation": False}},
+        api_base=api_base,
+        litellm_params=GenericLiteLLMParams(drop_params=True),
+        headers={},
+    )
+
+    assert {key: data[key] for key in data if key != "input"} == {"model": "gpt-5.4-mini", "stream": True}