From 7da34e8aed341b3368b14810d91052ebccb97117 Mon Sep 17 00:00:00 2001 From: Yassin Kortam Date: Fri, 21 Aug 2026 09:47:52 -0700 Subject: [PATCH] fix(proxy): make per-model budgets track spend, enforce, and report the same counter (#37736) Per-model budgets were three separate things pretending to be one. The enforcement check, the post-call increment and the info endpoints each derived their own cache key, so a budget could refuse traffic at 429 while /key/info reported zero usage, and a Bedrock model id never matched a budget keyed on the bare family name. /user/new echoed a model_max_budget back and stored an empty dict, and nothing enforced a user-scoped per-model budget at all. One owner now builds the counter key from the configured budget model, and enforcement, the increment and the info endpoints all read it. Bedrock ids resolve through the model-cost map. Auth carries the user's budget onto the token on every branch that reaches the spend hook, including JWT and auto-registration. Native passthrough attaches the three budget metadata keys its StandardLoggingUserAPIKeyMetadata does not carry, so /anthropic/... and /bedrock/... traffic is counted and capped like /v1/chat/completions. The dashboard gains the per-model budget editor it never had, on the key create, key edit and internal-user edit forms. It is read-only without an enterprise license, matching the write gate the proxy already enforces, and an untouched budget is left out of an update so an unrelated edit cannot trip that gate. The editor hydrates from either BudgetConfig spelling, since model_max_budget is a plain dict that the proxy stores exactly as the client sent it, and it carries through the fields it does not model. Without both, editing one model would drop another model row entirely and silently discard its tpm_limit and rpm_limit. /user/info refreshes its local copy of the user field by field after a save, so model_max_budget joins that list. Left out, a saved cap read back as the old one when the form was reopened, and clearing the row to recover would then wipe the value that had actually persisted. A zero-dollar cap is the strictest limit expressible, not the absence of one, so it is enforced rather than skipped on falsiness, spend exactly at the cap is refused the way every sibling budget check already refuses it, and a counter that was never written reads as zero spend rather than as unknown. The usage endpoints read every counter in one batched lookup, so a large model_max_budget cannot fan out into one concurrent cache call per configured model. Every auth path honours the same zero-cost skip flag, so none of them can refuse a free request that another serves. The custom-auth helper gains the flag it never had, which also changes its pre-existing key and end-user checks. The compaction summary gate checks the user scope alongside the key and end-user ones. This file propagates all three budgets into the summary subrequest, so enforcing only two let compaction increment a counter it could not be refused by. Custom auth attaches the user's budget to the token unconditionally, since the post-call spend hook reads it there: gating the attach on the same condition as enforcement left the counter uncharged whenever the request was not itself enforceable. An entry that will not validate is skipped rather than raised on, so one malformed scope cannot abort every other scope's increment or turn a config typo into a 500. The edit forms re-seed the budget editor when a different key or user is loaded. Its rows are seeded once and cannot re-read their own value prop, so without this a save wrote the previously loaded record's budgets onto the current one. Only the built-in provider pass-through routes carry the budget metadata. get_model_from_request deliberately resolves no model for a user-defined pass-through, since its body is forwarded verbatim and names an upstream model, so attaching there would charge a counter nothing on that route can refuse. --- .../context_management/editors/compact.py | 32 +- litellm/proxy/_types.py | 6 + litellm/proxy/auth/auth_utils.py | 4 +- litellm/proxy/auth/user_api_key_auth.py | 140 ++- .../proxy/hooks/model_max_budget_limiter.py | 584 +++++---- litellm/proxy/litellm_pre_call_utils.py | 2 + .../internal_user_endpoints.py | 21 +- .../key_management_endpoints.py | 84 +- .../pass_through_endpoints.py | 17 + ...test_unit_test_max_model_budget_limiter.py | 1054 ++++++++++++++--- .../test_user_api_key_auth.py | 670 ++++++++++- .../context_management/test_compact.py | 76 ++ .../test_internal_user_endpoints.py | 75 ++ .../test_key_management_endpoints.py | 104 +- .../test_pass_through_endpoints.py | 840 +++++-------- .../users/_components/BulkEditUsers.tsx | 3 + .../users/_components/user_edit_view.test.tsx | 121 +- .../users/_components/user_edit_view.tsx | 25 + .../user_info_view.integration.test.tsx | 44 +- .../_components/view_users/user_info_view.tsx | 8 + .../ModelMaxBudgetEditor.integration.test.tsx | 69 ++ .../ModelMaxBudgetEditor.test.ts | 140 +++ .../key_team_helpers/ModelMaxBudgetEditor.tsx | 233 ++++ .../components/key_team_helpers/key_list.tsx | 4 +- .../modelMaxBudgetPayload.test.ts | 71 ++ .../key_team_helpers/modelMaxBudgetPayload.ts | 44 + .../useModelMaxBudgetField.ts | 34 + .../key_team_helpers/useSeededState.ts | 26 + .../src/components/networking.tsx | 3 + .../organisms/createKeyPayload.test.ts | 25 + .../components/organisms/createKeyPayload.ts | 3 + .../organisms/create_key_button.tsx | 19 + .../templates/key_edit_view.test.tsx | 89 ++ .../components/templates/key_edit_view.tsx | 15 + ui/litellm-dashboard/src/lib/http/schema.d.ts | 12 + 35 files changed, 3656 insertions(+), 1041 deletions(-) create mode 100644 ui/litellm-dashboard/src/components/key_team_helpers/ModelMaxBudgetEditor.integration.test.tsx create mode 100644 ui/litellm-dashboard/src/components/key_team_helpers/ModelMaxBudgetEditor.test.ts create mode 100644 ui/litellm-dashboard/src/components/key_team_helpers/ModelMaxBudgetEditor.tsx create mode 100644 ui/litellm-dashboard/src/components/key_team_helpers/modelMaxBudgetPayload.test.ts create mode 100644 ui/litellm-dashboard/src/components/key_team_helpers/modelMaxBudgetPayload.ts create mode 100644 ui/litellm-dashboard/src/components/key_team_helpers/useModelMaxBudgetField.ts create mode 100644 ui/litellm-dashboard/src/components/key_team_helpers/useSeededState.ts diff --git a/litellm/llms/anthropic/experimental_pass_through/context_management/editors/compact.py b/litellm/llms/anthropic/experimental_pass_through/context_management/editors/compact.py index a7c462a8fb0..2a87afb5990 100644 --- a/litellm/llms/anthropic/experimental_pass_through/context_management/editors/compact.py +++ b/litellm/llms/anthropic/experimental_pass_through/context_management/editors/compact.py @@ -56,7 +56,7 @@ from ..result import PolyfillResult # so the summary's spend is attributed to the same scopes. The list mirrors the # fields populated by # ``LiteLLMProxyRequestSetup.add_user_api_key_auth_to_request_metadata``. -# ``user_api_key_model_max_budget`` / ``user_api_key_end_user_model_max_budget`` +# The three ``*_model_max_budget`` fields # are what ``_PROXY_VirtualKeyModelMaxBudgetLimiter`` reads post-call to update # the per-model spend caches, so without them the summary spend would never # count against the caller's model budget. ``user_api_key_end_user_id`` / @@ -76,6 +76,7 @@ _PROPAGATED_METADATA_KEYS: Final = ( "user_api_key_end_user_id", "user_api_end_user_max_budget", "user_api_key_model_max_budget", + "user_api_key_user_model_max_budget", "user_api_key_end_user_model_max_budget", "litellm_call_id", "litellm_parent_otel_span", @@ -317,10 +318,14 @@ async def _check_summary_model_budget( The summary subrequest never passes back through ``user_api_key_auth``, so without this gate a caller whose ``model_max_budget`` for ``context_management_summary_model`` is exhausted could keep consuming that - model via compaction. Mirrors the ``model_max_budget`` / - ``end_user_model_max_budget`` enforcement that ``user_api_key_auth`` runs for - the client-requested model. Returns True outside the proxy or when no + model via compaction. Mirrors the per-model budget enforcement that + ``user_api_key_auth`` runs for the client-requested model. Returns True outside the proxy or when no per-model budget is configured. + + All three scopes are checked because the summary's spend is charged to all + three: this file propagates the key, user and end-user budgets into the + subrequest's metadata, so enforcing only two of them would let compaction + increment a counter it can never be refused by. """ if user_api_key_auth is None: return True @@ -347,6 +352,25 @@ async def _check_summary_model_budget( ) return False + user_model_max_budget: Final = getattr(user_api_key_auth, "user_model_max_budget", None) + user_id: Final = getattr(user_api_key_auth, "user_id", None) + if isinstance(user_model_max_budget, dict) and user_model_max_budget and user_id is not None: + try: + await model_max_budget_limiter.is_user_within_model_budget( + user_id=user_id, + user_model_max_budget=user_model_max_budget, + model=summary_model, + ) + except litellm.BudgetExceededError: + return False + except Exception as e: # noqa: BLE001 # a budget gate denies on any failure, as the key and end-user scopes do + verbose_logger.warning( + "compact_20260112: unexpected error during user model-budget check for summary_model=%s; denying: %s", + summary_model, + e, + ) + return False + end_user_model_max_budget: Final = getattr(user_api_key_auth, "end_user_model_max_budget", None) end_user_id: Final = getattr(user_api_key_auth, "end_user_id", None) if isinstance(end_user_model_max_budget, dict) and end_user_model_max_budget and end_user_id is not None: diff --git a/litellm/proxy/_types.py b/litellm/proxy/_types.py index 00cbd13cfdc..e51a3138d2d 100644 --- a/litellm/proxy/_types.py +++ b/litellm/proxy/_types.py @@ -2805,6 +2805,10 @@ class UserAPIKeyAuth(LiteLLM_VerificationTokenView): # the expected response ob user_email: str | None = None user_spend: float | None = None user_max_budget: float | None = None + # Values stay `object` rather than BudgetConfig: this is the raw JSON column, + # and validating it here would make one malformed row fail auth outright. + # resolve_model_budget validates the single entry a request actually needs. + user_model_max_budget: dict[str, object] | None = None request_route: str | None = None is_session_token: bool = False # Server-only marker set exclusively by the MCP gateway admission path @@ -2982,6 +2986,8 @@ class UserInfoV2Response(LiteLLMPydanticObjectBase): sso_user_id: str | None = None teams: list[str] = [] # Just team IDs, not full team objects object_permission: LiteLLM_ObjectPermissionTable | None = None + model_max_budget: dict | None = None + model_max_budget_usage: dict | None = None from litellm.models.config import LiteLLM_Config as LiteLLM_Config # noqa: E402 diff --git a/litellm/proxy/auth/auth_utils.py b/litellm/proxy/auth/auth_utils.py index ce662ee0374..d04a71535ef 100644 --- a/litellm/proxy/auth/auth_utils.py +++ b/litellm/proxy/auth/auth_utils.py @@ -1801,7 +1801,7 @@ def _format_model_candidates( return candidates -def _request_dispatched_to_pass_through_endpoint(request: Request | None) -> bool: +def request_dispatched_to_pass_through_endpoint(request: Request | None) -> bool: """Whether FastAPI resolved this request to a user-defined pass-through handler. Reads the marker set by ``create_pass_through_route`` off the dispatched endpoint @@ -1842,7 +1842,7 @@ def get_model_from_request( and does not carry the marker. Built-in provider passthrough routes (``/vertex_ai``, ``/gemini``, ...) are separate handlers and keep model enforcement. """ - if _request_dispatched_to_pass_through_endpoint(request): + if request_dispatched_to_pass_through_endpoint(request): return None candidates: Final = _extract_model_candidates_from_request( diff --git a/litellm/proxy/auth/user_api_key_auth.py b/litellm/proxy/auth/user_api_key_auth.py index 99592d44f9b..fe4f1ee4ae5 100644 --- a/litellm/proxy/auth/user_api_key_auth.py +++ b/litellm/proxy/auth/user_api_key_auth.py @@ -11,6 +11,7 @@ import asyncio import fnmatch import re import secrets +from collections.abc import Mapping from datetime import datetime, timezone from typing import Any, Final, NamedTuple, Protocol, Union, cast @@ -186,6 +187,62 @@ class _KeyModelBudgetLimiter(Protocol): async def get_fallback_model_within_budget(self, user_api_key_dict: UserAPIKeyAuth, model: str) -> str | None: ... +class _UserModelBudgetLimiter(Protocol): + async def is_user_within_model_budget( + self, user_id: str, user_model_max_budget: Mapping[str, object], model: str + ) -> bool: ... + + +async def _read_user_model_max_budget( + user_id: str | None, + prisma_client: PrismaClient | None, + user_api_key_cache: UserApiKeyCache, + parent_otel_span: object, + proxy_logging_obj: ProxyLogging, +) -> dict | None: + """The user row's `model_max_budget`, or None when the row cannot be read. + + A user whose row is missing must not be refused: this is a budget lookup, + and the main auth path likewise treats an unreadable user as no user. + """ + if user_id is None or prisma_client is None: + return None + try: + user_obj: Final = await get_user_object( + user_id=user_id, + prisma_client=prisma_client, + user_api_key_cache=user_api_key_cache, + user_id_upsert=False, + parent_otel_span=parent_otel_span, # pyright: ignore[reportArgumentType] # Span is a runtime union, not usable in an annotation here + proxy_logging_obj=proxy_logging_obj, + ) + except Exception as e: # noqa: BLE001 # mirrors the main path's tolerance + verbose_logger.debug("Unable to read user for the per-model budget check: %s", e) + return None + return getattr(user_obj, "model_max_budget", None) + + +async def _check_user_model_budget( + valid_token: UserAPIKeyAuth, + model_max_budget_limiter: _UserModelBudgetLimiter, + models: list[str], +) -> None: + """Enforce the internal user's own `model_max_budget` across the request's models. + + Separate from the key check: a user's per-model budget caps every key they + own, so a caller cannot escape it by minting another key. + """ + user_model_max_budget: Final = valid_token.user_model_max_budget + if valid_token.user_id is None or not isinstance(user_model_max_budget, Mapping) or not user_model_max_budget: + return + for model_name in models: + await model_max_budget_limiter.is_user_within_model_budget( + user_id=valid_token.user_id, + user_model_max_budget=user_model_max_budget, + model=model_name, + ) + + async def _check_key_model_budget_with_fallback( valid_token: UserAPIKeyAuth, model_max_budget_limiter: _KeyModelBudgetLimiter, @@ -1390,6 +1447,7 @@ async def _user_api_key_auth_builder( end_user_id=end_user_id, user_tpm_limit=(user_object.tpm_limit if user_object is not None else None), user_rpm_limit=(user_object.rpm_limit if user_object is not None else None), + user_model_max_budget=(user_object.model_max_budget if user_object is not None else None), team_member_rpm_limit=( team_membership.safe_get_team_member_rpm_limit() if team_membership is not None else None ), @@ -1427,6 +1485,13 @@ async def _user_api_key_auth_builder( if auto_registered is not None: auto_registered.jwt_claims = jwt_claims auto_registered.user_email = user_email + # The auto-registered token is built from the new key's + # columns, which carry no user budget. Carry over the + # already-loaded user row rather than re-reading it, or + # the budget check below has nothing to enforce. + auto_registered.user_model_max_budget = ( + user_object.model_max_budget if user_object is not None else None + ) valid_token = auto_registered api_key = valid_token.token or "" @@ -1458,6 +1523,28 @@ async def _user_api_key_auth_builder( valid_token.project_metadata = _jwt_project_obj.metadata valid_token.project_alias = _jwt_project_obj.project_alias + # JWT auth returns here rather than falling through to the + # virtual-key checks below, so the user's per-model budget + # has to be enforced on this path too. Without it the + # post-call increment still charges the counter and nothing + # ever reads it, which is worse than not tracking at all. + # Guarded by the same flag the virtual-key path uses, or a + # zero-cost model would be refused here and allowed there, + # while the log above claims all budget checks were skipped. + if not skip_budget_checks: + await _check_user_model_budget( + valid_token=cast(UserAPIKeyAuth, valid_token), + model_max_budget_limiter=model_max_budget_limiter, + models=_get_model_names_for_budget_checks( + model=_get_model_from_request_context( + request_data=request_data, + route=route, + request=request, + llm_router=llm_router, + ) + ), + ) + return cast(UserAPIKeyAuth, valid_token) #### ELSE #### @@ -1811,6 +1898,12 @@ async def _user_api_key_auth_builder( ) user_obj = None + if user_obj is not None: + # The joint verification-token view carries the key's columns only, so the + # user's own per-model budget reaches enforcement and the post-call + # increment through the row fetched here. + valid_token.user_model_max_budget = user_obj.model_max_budget + if ( user_obj is not None and isinstance(user_obj.metadata, dict) @@ -1974,6 +2067,14 @@ async def _user_api_key_auth_builder( ) current_models = _get_model_names_for_budget_checks(model=current_model) + # Check 5a. Internal user model_max_budget + if current_models: + await _check_user_model_budget( + valid_token=valid_token, + model_max_budget_limiter=model_max_budget_limiter, + models=current_models, + ) + # Check 5b. End-user model max budget end_user_mmb: Final = valid_token.end_user_model_max_budget if ( @@ -2757,6 +2858,7 @@ async def _return_user_api_key_auth_obj( user_email=user_obj.user_email, user_spend=getattr(user_obj, "spend", None), user_max_budget=getattr(user_obj, "max_budget", None), + user_model_max_budget=getattr(user_obj, "model_max_budget", None), ) if user_obj is not None and _is_user_proxy_admin(user_obj=user_obj): user_api_key_kwargs.update( @@ -3020,10 +3122,21 @@ async def _run_post_custom_auth_checks( ) current_models = _get_model_names_for_budget_checks(model=current_model) + # A zero-cost model cannot move any counter, so refusing it means refusing on + # spend some other model accrued. The JWT and virtual-key paths already skip + # every budget check for these; this path did not, so the same request could + # be refused under custom auth and served under the other two. + skip_budget_checks: Final = ( + _is_model_cost_zero(model=current_model, llm_router=llm_router) + if current_model is not None and llm_router is not None + else False + ) + # 3. Check key-level model_max_budget max_budget_per_model: Final = valid_token.model_max_budget if ( - max_budget_per_model is not None + not skip_budget_checks + and max_budget_per_model is not None and isinstance(max_budget_per_model, dict) and len(max_budget_per_model) > 0 and current_models @@ -3050,10 +3163,33 @@ async def _run_post_custom_auth_checks( ) current_models = _get_model_names_for_budget_checks(model=current_model) + # 3b. Attach and check the internal user's model_max_budget. + # Custom auth builds its own token, so unlike the main path nothing has + # loaded the user row yet. The attach is unconditional because the post-call + # spend hook reads this field off the token: gating it on the same condition + # as enforcement would leave the user's counter uncharged whenever this + # request was not itself enforceable, which is the untracked-spend bug this + # PR exists to fix. + user_budget: Final = await _read_user_model_max_budget( + user_id=valid_token.user_id, + prisma_client=prisma_client, + user_api_key_cache=user_api_key_cache, + parent_otel_span=parent_otel_span, + proxy_logging_obj=proxy_logging_obj, + ) + valid_token.user_model_max_budget = user_budget # rebind-ok: the spend hook reads it off this token + if not skip_budget_checks and current_models: + await _check_user_model_budget( + valid_token=valid_token, + model_max_budget_limiter=model_max_budget_limiter, + models=current_models, + ) + # 4. Check end-user model_max_budget end_user_mmb: Final = valid_token.end_user_model_max_budget if ( - end_user_mmb is not None + not skip_budget_checks + and end_user_mmb is not None and isinstance(end_user_mmb, dict) and len(end_user_mmb) > 0 and current_models diff --git a/litellm/proxy/hooks/model_max_budget_limiter.py b/litellm/proxy/hooks/model_max_budget_limiter.py index 215969ef899..c5d10b2749b 100644 --- a/litellm/proxy/hooks/model_max_budget_limiter.py +++ b/litellm/proxy/hooks/model_max_budget_limiter.py @@ -1,21 +1,253 @@ import json +import time +from collections.abc import Iterable, Mapping +from dataclasses import dataclass +from types import MappingProxyType from typing import Final import litellm from litellm._logging import verbose_proxy_logger from litellm.caching.caching import DualCache from litellm.integrations.custom_logger import Span +from litellm.litellm_core_utils.duration_parser import duration_in_seconds +from litellm.llms.bedrock.common_utils import get_bedrock_base_model from litellm.proxy._types import Litellm_EntityType, UserAPIKeyAuth from litellm.router_strategy.budget_limiter import RouterBudgetLimiting from litellm.types.llms.openai import AllMessageValues -from litellm.types.utils import ( - BudgetConfig, - GenericBudgetConfigType, - StandardLoggingPayload, -) +from litellm.types.utils import BudgetConfig, StandardLoggingPayload VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX: Final = "virtual_key_spend" END_USER_SPEND_CACHE_KEY_PREFIX: Final = "end_user_model_spend" +USER_SPEND_CACHE_KEY_PREFIX: Final = "user_model_spend" + +_SPEND_CACHE_KEY_PREFIXES: Final = MappingProxyType( + { + Litellm_EntityType.KEY: VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX, + Litellm_EntityType.USER: USER_SPEND_CACHE_KEY_PREFIX, + Litellm_EntityType.END_USER: END_USER_SPEND_CACHE_KEY_PREFIX, + } +) + +_LEGACY_REQUEST_MODEL_SCOPES: Final = frozenset({Litellm_EntityType.KEY, Litellm_EntityType.END_USER}) + +_PROCESS_STARTED_AT: Final = time.monotonic() + +_BUDGET_START_TIME_KEY_PREFIXES: Final = MappingProxyType( + { + Litellm_EntityType.KEY: "virtual_key_budget_start_time", + Litellm_EntityType.USER: "user_model_budget_start_time", + Litellm_EntityType.END_USER: "end_user_budget_start_time", + } +) + + +@dataclass(frozen=True, slots=True) +class ResolvedModelBudget: + """The `model_max_budget` entry a request resolved to. + + ``budget_model`` is the key as the operator configured it, not the model + name on the request. Every counter is keyed on it so enforcement, the + post-call increment and the `/key/info` + `/user/info` usage reads cannot + disagree about which counter a request belongs to. + """ + + budget_model: str + budget_config: BudgetConfig + + +def model_budget_spend_cache_key( + entity_type: Litellm_EntityType, + entity_id: str | None, + budget_model: str, + budget_duration: str | None, +) -> str: + """Sole owner of the per-model spend counter key, shared by its writer and all of its readers.""" + return f"{_SPEND_CACHE_KEY_PREFIXES[entity_type]}:{entity_id}:{budget_model}:{budget_duration}" + + +def _legacy_request_model_spend_cache_key( + entity_type: Litellm_EntityType, + entity_id: str | None, + model: str, + resolved: ResolvedModelBudget, +) -> str | None: + """The counter this request was billed to before the budget model owned the key, or None. + + Upgrading proxies carry live counters keyed on the model as REQUESTED + (`openai/gpt-4`) rather than as configured (`gpt-4`), and those were the + counters the previous version enforced on. Nothing writes that spelling once + this version is running, so the pre-upgrade and post-upgrade counters hold + disjoint halves of one window and adding them is the window's real spend. + + Only the key and end-user scopes ever had one. The user scope is introduced + by this change, so it has no counter to carry. + + The carry stops one budget window after start-up, because a legacy counter + belongs to a window that was already open when this process replaced the one + writing it. Past that point the lookup could only ever miss. + """ + budget_duration: Final = resolved.budget_config.budget_duration + if entity_type not in _LEGACY_REQUEST_MODEL_SCOPES or budget_duration is None: + return None + if time.monotonic() - _PROCESS_STARTED_AT >= duration_in_seconds(budget_duration): + return None + return model_budget_spend_cache_key( + entity_type=entity_type, + entity_id=entity_id, + budget_model=model, + budget_duration=budget_duration, + ) + + +def model_budget_start_time_cache_key( + entity_type: Litellm_EntityType, + entity_id: str | None, + budget_model: str, + budget_duration: str | None, +) -> str: + """Window start for one (entity, budget model) pair. + + Scoped per budget model because an entity may budget two models over + different periods, and a shared start time lets the shorter period restart + the longer one's window. + """ + return f"{_BUDGET_START_TIME_KEY_PREFIXES[entity_type]}:{entity_id}:{budget_model}:{budget_duration}" + + +def resolve_model_budget(model: str, model_max_budget: Mapping[str, object]) -> ResolvedModelBudget | None: + """Find the `model_max_budget` entry that governs `model`, or None.""" + for candidate in _budget_model_candidates(model): + raw_budget_config = model_max_budget.get(candidate) + if raw_budget_config is None: + continue + if (budget_config := _usable_budget_config(raw_budget_config)) is None: + # An entry that will not validate cannot be keyed, so it cannot be + # enforced or incremented. Skip to the next candidate rather than + # raising: raising would abort every other scope's increment and turn + # a config typo into a 500, and stopping here would let one malformed + # specific entry disable a perfectly good bare-family budget beside + # it. The candidate chain already falls through an ABSENT entry, and + # an unparseable one is indistinguishable from absent to enforcement. + # `validate_model_max_budget` rejects these on the write path, so + # reaching here means config.yaml or a direct DB edit. + verbose_proxy_logger.warning( + "Ignoring unusable model_max_budget entry for %s; it cannot be enforced or tracked", + candidate, + ) + continue + return ResolvedModelBudget(budget_model=candidate, budget_config=budget_config) + return None + + +def _budget_model_candidates(model: str) -> tuple[str, ...]: + """Names a budget may be configured under for a request on `model`, most specific first. + + Beyond the model as sent, a budget may be keyed on the model without its + ``{custom_llm_provider}/`` prefix (``gpt-4o`` governs ``openai/gpt-4o``), on + the Bedrock base model (``anthropic.claude-opus-4-8`` governs the + cross-region ``us.anthropic.claude-opus-4-8``), or on the bare family name + that Bedrock id shares with its direct-provider twin (``claude-opus-4-8``). + """ + return tuple(dict.fromkeys((model, model.split("/")[-1], *_bedrock_candidates(model)))) + + +def _bedrock_candidates(model: str) -> tuple[str, ...]: + """Bedrock-only candidates, empty unless litellm prices `model` as a Bedrock model. + + Gating on the cost map rather than on a vendor allowlist is what makes + splitting the leading dotted segment safe: most dotted model ids are not + Bedrock ids at all (``azure/gpt-4.1``, ``gpt-image-1.5``), and splitting one + of those would produce a garbage candidate. + """ + base_model: Final = get_bedrock_base_model(model) + cost_entry: Final = litellm.model_cost.get(base_model) + if not isinstance(cost_entry, dict) or not str(cost_entry.get("litellm_provider", "")).startswith("bedrock"): + return () + _, _, without_vendor = base_model.partition(".") + return (base_model, without_vendor) if without_vendor else (base_model,) + + +async def build_model_max_budget_usage( + entity_type: Litellm_EntityType, + entity_id: str | None, + model_max_budget: Mapping[str, object] | None, + cache: DualCache | None, +) -> dict[str, dict[str, object]]: + """Current-window spend per configured budget model, as `/key/info` and `/user/info` report it. + + `cache` must be the DualCache the limiter writes the counters to; callers + read it off the limiter rather than re-deriving it, so a scope that is being + blocked can never report zero usage. + """ + if cache is None or entity_id is None or not model_max_budget: + return {} + + budgets: Final = tuple( + (budget_model, budget_config) + for budget_model, raw_budget_config in model_max_budget.items() + for budget_config in (_usable_budget_config(raw_budget_config),) + if budget_config is not None + ) + if not budgets: + return {} + spend_keys: Final = tuple( + model_budget_spend_cache_key( + entity_type=entity_type, + entity_id=entity_id, + budget_model=budget_model, + budget_duration=budget_config.budget_duration, + ) + for budget_model, budget_config in budgets + ) + batched: Final = await cache.async_batch_get_cache( + keys=list(spend_keys) # mutable-ok: async_batch_get_cache annotates keys as list, so one must exist here + ) + # async_batch_get_cache returns None if it fails internally, and its result is + # index-aligned with `keys` otherwise. An unusable result reads as a miss, + # which is what a never-written counter already reads as. + current_spends: Final = ( + tuple(batched) if isinstance(batched, list) and len(batched) == len(budgets) else (None,) * len(budgets) + ) + return { + budget_model: { + "current_spend": round(_as_spend(current_spend), 4), + "budget_limit": budget_config.max_budget, + "time_period": budget_config.budget_duration, + } + for (budget_model, budget_config), current_spend in zip(budgets, current_spends, strict=True) + } + + +def _usable_budget_config(raw_budget_config: object) -> BudgetConfig | None: + try: + budget_config: Final = BudgetConfig.model_validate(raw_budget_config) + if budget_config.budget_duration is None: + return None + duration_in_seconds(budget_config.budget_duration) + except Exception: # noqa: BLE001 # a malformed entry must not fail the whole report + return None + return budget_config + + +def _as_spend(current_spend: object) -> float: + try: + return float(current_spend or 0.0) # pyright: ignore[reportArgumentType] # non-numeric falls to the except + except (TypeError, ValueError): + return 0.0 + + +def _resolve_entity_model_budgets( + model: str, + entity_budgets: Iterable[tuple[Litellm_EntityType, str | None, object]], +) -> tuple[tuple[Litellm_EntityType, str, ResolvedModelBudget], ...]: + """Drop the scopes that do not budget `model`, keeping only what can be incremented.""" + return tuple( + (entity_type, entity_id, resolved) + for entity_type, entity_id, model_max_budget in entity_budgets + if entity_id is not None and isinstance(model_max_budget, Mapping) and model_max_budget + for resolved in (resolve_model_budget(model=model, model_max_budget=model_max_budget),) + if resolved is not None and resolved.budget_config.budget_duration is not None + ) class _PROXY_VirtualKeyModelMaxBudgetLimiter(RouterBudgetLimiting): @@ -41,47 +273,17 @@ class _PROXY_VirtualKeyModelMaxBudgetLimiter(RouterBudgetLimiting): Raises: BudgetExceededError: If the user_api_key_dict has exceeded the model budget """ - _model_max_budget: Final = user_api_key_dict.model_max_budget - internal_model_max_budget: Final[GenericBudgetConfigType] = {} - - for _model, _budget_info in _model_max_budget.items(): - internal_model_max_budget[_model] = BudgetConfig(**_budget_info) - - verbose_proxy_logger.debug( - "internal_model_max_budget %s", - json.dumps(internal_model_max_budget, indent=4, default=str), + return await self._is_entity_within_model_budget( + entity_type=Litellm_EntityType.KEY, + entity_id=user_api_key_dict.token, + model_max_budget=user_api_key_dict.model_max_budget, + model=model, + exceeded_message=( + f"LiteLLM Virtual Key: {user_api_key_dict.token}, key_alias: {user_api_key_dict.key_alias}, " + f"exceeded budget for model={model}" + ), ) - # check if current model is in internal_model_max_budget - _current_model_budget_info: Final = self._get_request_model_budget_config( - model=model, internal_model_max_budget=internal_model_max_budget - ) - if _current_model_budget_info is None: - verbose_proxy_logger.debug("Model %s not found in internal_model_max_budget", model) - return True - - # check if current model is within budget - if _current_model_budget_info.max_budget and _current_model_budget_info.max_budget > 0: - _current_spend: Final = await self._get_virtual_key_spend_for_model( - user_api_key_hash=user_api_key_dict.token, - model=model, - key_budget_config=_current_model_budget_info, - ) - if ( - _current_spend is not None - and _current_model_budget_info.max_budget is not None - and _current_spend > _current_model_budget_info.max_budget - ): - raise litellm.BudgetExceededError( - message=f"LiteLLM Virtual Key: {user_api_key_dict.token}, key_alias: {user_api_key_dict.key_alias}, exceeded budget for model={model}", - current_cost=_current_spend, - max_budget=_current_model_budget_info.max_budget, - entity_type=Litellm_EntityType.KEY.value, - entity_id=user_api_key_dict.token, - ) - - return True - async def get_fallback_model_within_budget( self, user_api_key_dict: UserAPIKeyAuth, @@ -96,10 +298,30 @@ class _PROXY_VirtualKeyModelMaxBudgetLimiter(RouterBudgetLimiting): continue return None + async def is_user_within_model_budget( + self, + user_id: str, + user_model_max_budget: Mapping[str, object], + model: str, + ) -> bool: + """ + Check if the internal user is within the model budget + + Raises: + BudgetExceededError: If the user has exceeded the model budget + """ + return await self._is_entity_within_model_budget( + entity_type=Litellm_EntityType.USER, + entity_id=user_id, + model_max_budget=user_model_max_budget, + model=model, + exceeded_message=f"LiteLLM User: {user_id}, exceeded budget for model={model}", + ) + async def is_end_user_within_model_budget( self, end_user_id: str, - end_user_model_max_budget: dict, + end_user_model_max_budget: Mapping[str, object], model: str, ) -> bool: """ @@ -108,116 +330,81 @@ class _PROXY_VirtualKeyModelMaxBudgetLimiter(RouterBudgetLimiting): Raises: BudgetExceededError: If the end_user has exceeded the model budget """ - internal_model_max_budget: Final[GenericBudgetConfigType] = {} - - for _model, _budget_info in end_user_model_max_budget.items(): - internal_model_max_budget[_model] = BudgetConfig(**_budget_info) - - verbose_proxy_logger.debug( - "end_user internal_model_max_budget %s", - json.dumps(internal_model_max_budget, indent=4, default=str), + return await self._is_entity_within_model_budget( + entity_type=Litellm_EntityType.END_USER, + entity_id=end_user_id, + model_max_budget=end_user_model_max_budget, + model=model, + exceeded_message=f"LiteLLM End User: {end_user_id}, exceeded budget for model={model}", ) - # check if current model is in internal_model_max_budget - _current_model_budget_info: Final = self._get_request_model_budget_config( - model=model, internal_model_max_budget=internal_model_max_budget - ) - if _current_model_budget_info is None: - verbose_proxy_logger.debug("Model %s not found in end_user_model_max_budget", model) + async def _is_entity_within_model_budget( + self, + entity_type: Litellm_EntityType, + entity_id: str | None, + model_max_budget: Mapping[str, object] | None, + model: str, + exceeded_message: str, + ) -> bool: + if not model_max_budget: + return True + resolved: Final = resolve_model_budget(model=model, model_max_budget=model_max_budget) + if resolved is None: + verbose_proxy_logger.debug("Model %s not found in %s model_max_budget", model, entity_type.value) return True - # check if current model is within budget - if _current_model_budget_info.max_budget and _current_model_budget_info.max_budget > 0: - _current_spend: Final = await self._get_end_user_spend_for_model( - end_user_id=end_user_id, - model=model, - key_budget_config=_current_model_budget_info, - ) - if ( - _current_spend is not None - and _current_model_budget_info.max_budget is not None - and _current_spend > _current_model_budget_info.max_budget - ): - raise litellm.BudgetExceededError( - message=f"LiteLLM End User: {end_user_id}, exceeded budget for model={model}", - current_cost=_current_spend, - max_budget=_current_model_budget_info.max_budget, - entity_type=Litellm_EntityType.END_USER.value, - entity_id=end_user_id, - ) + max_budget: Final = resolved.budget_config.max_budget + if max_budget is None or max_budget < 0: + return True + current_spend: Final = await self._get_spend_for_model_budget( + entity_type=entity_type, + entity_id=entity_id, + model=model, + resolved=resolved, + ) + if current_spend >= max_budget: + raise litellm.BudgetExceededError( + message=exceeded_message, + current_cost=current_spend, + max_budget=max_budget, + entity_type=entity_type.value, + entity_id=entity_id, + ) return True - async def _get_end_user_spend_for_model( + async def _get_spend_for_model_budget( self, - end_user_id: str, + entity_type: Litellm_EntityType, + entity_id: str | None, model: str, - key_budget_config: BudgetConfig, - ) -> float | None: - # 1. model: directly look up `model` - end_user_model_spend_cache_key = ( - f"{END_USER_SPEND_CACHE_KEY_PREFIX}:{end_user_id}:{model}:{key_budget_config.budget_duration}" - ) - _current_spend = await self.dual_cache.async_get_cache( - key=end_user_model_spend_cache_key, - ) + resolved: ResolvedModelBudget, + ) -> float: + """Spend charged to this budget in the current window, legacy counter included. - if _current_spend is None: - # 2. If 1, does not exist, check if passed as {custom_llm_provider}/model - end_user_model_spend_cache_key = f"{END_USER_SPEND_CACHE_KEY_PREFIX}:{end_user_id}:{self._get_model_without_custom_llm_provider(model)}:{key_budget_config.budget_duration}" - _current_spend = await self.dual_cache.async_get_cache( - key=end_user_model_spend_cache_key, - ) - return _current_spend - - async def _get_virtual_key_spend_for_model( - self, - user_api_key_hash: str | None, - model: str, - key_budget_config: BudgetConfig, - ) -> float | None: + A counter that was never written is zero spend, not unknown spend. The + distinction only shows up at a zero-dollar cap, where skipping the + comparison would let the strictest possible limit admit every request. """ - Get the current spend for a virtual key for a model - - Lookup model in this order: - 1. model: directly look up `model` - 2. If 1, does not exist, check if passed as {custom_llm_provider}/model - """ - - # 1. model: directly look up `model` - virtual_key_model_spend_cache_key = ( - f"{VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX}:{user_api_key_hash}:{model}:{key_budget_config.budget_duration}" + spend_key: Final = model_budget_spend_cache_key( + entity_type=entity_type, + entity_id=entity_id, + budget_model=resolved.budget_model, + budget_duration=resolved.budget_config.budget_duration, ) - _current_spend = await self.dual_cache.async_get_cache( - key=virtual_key_model_spend_cache_key, + legacy_spend_key: Final = _legacy_request_model_spend_cache_key( + entity_type=entity_type, + entity_id=entity_id, + model=model, + resolved=resolved, ) + current_spend: Final = _as_spend(await self._cached_spend(spend_key)) + if legacy_spend_key is None or legacy_spend_key == spend_key: + return current_spend + return current_spend + _as_spend(await self._cached_spend(legacy_spend_key)) - if _current_spend is None: - # 2. If 1, does not exist, check if passed as {custom_llm_provider}/model - # if "/" in model, remove first part before "/" - eg. openai/o1-preview -> o1-preview - virtual_key_model_spend_cache_key = f"{VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX}:{user_api_key_hash}:{self._get_model_without_custom_llm_provider(model)}:{key_budget_config.budget_duration}" - _current_spend = await self.dual_cache.async_get_cache( - key=virtual_key_model_spend_cache_key, - ) - return _current_spend - - def _get_request_model_budget_config( - self, model: str, internal_model_max_budget: GenericBudgetConfigType - ) -> BudgetConfig | None: - """ - Get the budget config for the request model - - 1. Check if `model` is in `internal_model_max_budget` - 2. If not, check if `model` without custom llm provider is in `internal_model_max_budget` - """ - return internal_model_max_budget.get(model, None) or internal_model_max_budget.get( - self._get_model_without_custom_llm_provider(model), None - ) - - def _get_model_without_custom_llm_provider(self, model: str) -> str: - if "/" in model: - return model.split("/")[-1] - return model + async def _cached_spend(self, spend_key: str) -> float | None: + return await self.dual_cache.async_get_cache(key=spend_key) async def async_filter_deployments( self, @@ -245,80 +432,63 @@ class _PROXY_VirtualKeyModelMaxBudgetLimiter(RouterBudgetLimiting): _litellm_params: Final[dict] = kwargs.get("litellm_params", {}) or {} _metadata: Final[dict] = _litellm_params.get("metadata", {}) or {} - user_api_key_model_max_budget: Final[dict | None] = _metadata.get("user_api_key_model_max_budget", None) - user_api_key_end_user_model_max_budget: Final[dict | None] = _metadata.get( - "user_api_key_end_user_model_max_budget", None - ) - if (user_api_key_model_max_budget is None or len(user_api_key_model_max_budget) == 0) and ( - user_api_key_end_user_model_max_budget is None or len(user_api_key_end_user_model_max_budget) == 0 - ): - verbose_proxy_logger.debug( - "Not running _PROXY_VirtualKeyModelMaxBudgetLimiter.async_log_success_event because user_api_key_model_max_budget and user_api_key_end_user_model_max_budget are None or empty." - ) - return + payload_metadata: Final = standard_logging_payload.get("metadata") or {} - response_cost: Final[float] = standard_logging_payload.get("response_cost", 0) # Use model_group (the user-facing model alias, e.g. "gpt-4o") when - # available. The enforcement path (is_key_within_model_budget) receives - # the model name from request_data["model"] which is the model group - # alias, so the spend tracking cache key must use the same name. - # Falling back to the deployment-level "model" field preserves - # behaviour for non-proxy or non-router deployments where model_group - # is None. + # available. The enforcement path receives the model name from + # request_data["model"] which is the model group alias, so the spend + # tracking cache key must resolve from the same name. Falling back to + # the deployment-level "model" field preserves behaviour for non-proxy + # or non-router deployments where model_group is None. model: Final = standard_logging_payload.get("model_group") or standard_logging_payload.get("model") - virtual_key: Final = standard_logging_payload.get("metadata", {}).get("user_api_key_hash") - end_user_id = standard_logging_payload.get("end_user") or standard_logging_payload.get("metadata", {}).get( - "user_api_key_end_user_id" - ) - if model is None: return - if ( - virtual_key is not None - and user_api_key_model_max_budget is not None - and len(user_api_key_model_max_budget) > 0 - ): - internal_model_max_budget: GenericBudgetConfigType = {} - for _model, _budget_info in user_api_key_model_max_budget.items(): - internal_model_max_budget[_model] = BudgetConfig(**_budget_info) - key_budget_config = self._get_request_model_budget_config( - model=model, internal_model_max_budget=internal_model_max_budget - ) - if key_budget_config is not None and key_budget_config.budget_duration: - virtual_spend_key: Final = ( - f"{VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX}:{virtual_key}:{model}:{key_budget_config.budget_duration}" - ) - virtual_start_time_key: Final = f"virtual_key_budget_start_time:{virtual_key}" - await self._increment_spend_for_key( - budget_config=key_budget_config, - spend_key=virtual_spend_key, - start_time_key=virtual_start_time_key, - response_cost=response_cost, - ) + response_cost: Final[float] = standard_logging_payload.get("response_cost", 0) + entity_budgets: Final = ( + ( + Litellm_EntityType.KEY, + payload_metadata.get("user_api_key_hash"), + _metadata.get("user_api_key_model_max_budget"), + ), + ( + Litellm_EntityType.USER, + payload_metadata.get("user_api_key_user_id"), + _metadata.get("user_api_key_user_model_max_budget"), + ), + ( + Litellm_EntityType.END_USER, + standard_logging_payload.get("end_user") or payload_metadata.get("user_api_key_end_user_id"), + _metadata.get("user_api_key_end_user_model_max_budget"), + ), + ) - if ( - end_user_id is not None - and user_api_key_end_user_model_max_budget is not None - and len(user_api_key_end_user_model_max_budget) > 0 - ): - internal_model_max_budget: GenericBudgetConfigType = {} - for _model, _budget_info in user_api_key_end_user_model_max_budget.items(): - internal_model_max_budget[_model] = BudgetConfig(**_budget_info) - key_budget_config = self._get_request_model_budget_config( - model=model, internal_model_max_budget=internal_model_max_budget + resolved_budgets: Final = _resolve_entity_model_budgets(model=model, entity_budgets=entity_budgets) + if not resolved_budgets: + verbose_proxy_logger.debug( + "Not running _PROXY_VirtualKeyModelMaxBudgetLimiter.async_log_success_event: " + "no key, user or end-user model_max_budget covers model=%s", + model, + ) + return + + for entity_type, entity_id, resolved in resolved_budgets: + await self._increment_spend_for_key( + budget_config=resolved.budget_config, + spend_key=model_budget_spend_cache_key( + entity_type=entity_type, + entity_id=entity_id, + budget_model=resolved.budget_model, + budget_duration=resolved.budget_config.budget_duration, + ), + start_time_key=model_budget_start_time_cache_key( + entity_type=entity_type, + entity_id=entity_id, + budget_model=resolved.budget_model, + budget_duration=resolved.budget_config.budget_duration, + ), + response_cost=response_cost, ) - if key_budget_config is not None and key_budget_config.budget_duration: - end_user_spend_key: Final = ( - f"{END_USER_SPEND_CACHE_KEY_PREFIX}:{end_user_id}:{model}:{key_budget_config.budget_duration}" - ) - end_user_start_time_key: Final = f"end_user_budget_start_time:{end_user_id}" - await self._increment_spend_for_key( - budget_config=key_budget_config, - spend_key=end_user_spend_key, - start_time_key=end_user_start_time_key, - response_cost=response_cost, - ) if self.dual_cache.redis_cache is not None: await self._push_in_memory_increments_to_redis() diff --git a/litellm/proxy/litellm_pre_call_utils.py b/litellm/proxy/litellm_pre_call_utils.py index 2ec5c34958c..1541b8acfdc 100644 --- a/litellm/proxy/litellm_pre_call_utils.py +++ b/litellm/proxy/litellm_pre_call_utils.py @@ -1943,6 +1943,8 @@ async def add_litellm_data_to_request( # Follow same pattern as team and API key budgets data[_metadata_variable_name]["user_api_key_user_spend"] = user_api_key_dict.user_spend data[_metadata_variable_name]["user_api_key_user_max_budget"] = user_api_key_dict.user_max_budget + user_model_budget: Final = user_api_key_dict.user_model_max_budget + data[_metadata_variable_name]["user_api_key_user_model_max_budget"] = user_model_budget # rebind-ok: out-param data[_metadata_variable_name]["user_api_key_metadata"] = strip_callback_config(user_api_key_dict.metadata) data[_metadata_variable_name]["user_api_key_team_metadata"] = strip_callback_config(user_api_key_dict.team_metadata) diff --git a/litellm/proxy/management_endpoints/internal_user_endpoints.py b/litellm/proxy/management_endpoints/internal_user_endpoints.py index 9c725c54d08..c2f5b8eeb8b 100644 --- a/litellm/proxy/management_endpoints/internal_user_endpoints.py +++ b/litellm/proxy/management_endpoints/internal_user_endpoints.py @@ -32,6 +32,7 @@ from litellm.proxy.common_utils.user_api_key_cache import ( object_permission_cache_key, user_object_permission_id_cache_key, ) +from litellm.proxy.hooks.model_max_budget_limiter import build_model_max_budget_usage from litellm.proxy.hooks.user_management_event_hooks import UserManagementEventHooks from litellm.proxy.management_endpoints.common_daily_activity import ( DailySpendRecord, @@ -817,6 +818,7 @@ def _build_user_info_response( keys: list[LiteLLM_VerificationToken] | None, team_list: list[TeamListResponseObject], teams_1: list[TeamListResponseObject] | None, + model_max_budget_usage: dict[str, dict[str, object]] | None = None, ) -> UserInfoResponse: """Create UserInfoResponse while filtering sensitive fields.""" if user_info is None and keys is not None: @@ -830,6 +832,8 @@ def _build_user_info_response( if isinstance(_user_info, dict): _user_info.pop("password", None) _user_info["metadata"] = _redact_scim_enterprise_metadata(_user_info.get("metadata")) + if model_max_budget_usage is not None: + _user_info["model_max_budget_usage"] = model_max_budget_usage return UserInfoResponse( user_id=user_id, @@ -864,7 +868,7 @@ async def user_info( --header 'Authorization: Bearer sk-1234' ``` """ - from litellm.proxy.proxy_server import prisma_client + from litellm.proxy.proxy_server import model_max_budget_limiter, prisma_client try: user_id = _normalize_user_info_user_id(request=request, user_id=user_id) @@ -910,6 +914,12 @@ async def user_info( keys=keys, team_list=team_list, teams_1=teams_1, + model_max_budget_usage=await build_model_max_budget_usage( + entity_type=Litellm_EntityType.USER, + entity_id=user_id, + model_max_budget=getattr(user_info, "model_max_budget", None), + cache=model_max_budget_limiter.dual_cache, + ), ) return response_data @@ -1007,7 +1017,7 @@ async def user_info_v2( --header 'Authorization: Bearer sk-1234' ``` """ - from litellm.proxy.proxy_server import prisma_client + from litellm.proxy.proxy_server import model_max_budget_limiter, prisma_client try: if prisma_client is None: @@ -1062,6 +1072,13 @@ async def user_info_v2( sso_user_id=user_data.get("sso_user_id"), teams=user_data.get("teams") or [], object_permission=user_data.get("object_permission"), + model_max_budget=user_data.get("model_max_budget"), + model_max_budget_usage=await build_model_max_budget_usage( + entity_type=Litellm_EntityType.USER, + entity_id=user_data.get("user_id", user_id), + model_max_budget=user_data.get("model_max_budget"), + cache=model_max_budget_limiter.dual_cache, + ), ) except Exception as e: verbose_proxy_logger.exception("litellm.proxy.proxy_server.user_info_v2(): Exception occured - %s", e) diff --git a/litellm/proxy/management_endpoints/key_management_endpoints.py b/litellm/proxy/management_endpoints/key_management_endpoints.py index 71218d6114b..bf42aeeec05 100644 --- a/litellm/proxy/management_endpoints/key_management_endpoints.py +++ b/litellm/proxy/management_endpoints/key_management_endpoints.py @@ -29,6 +29,7 @@ from fastapi import APIRouter, Depends, Header, HTTPException, Query, Request, s import litellm from litellm._logging import verbose_proxy_logger from litellm._uuid import uuid +from litellm.caching.dual_cache import DualCache from litellm.constants import ( LENGTH_OF_LITELLM_GENERATED_KEY, LITELLM_PROXY_ADMIN_NAME, @@ -47,7 +48,7 @@ from litellm.proxy._experimental.mcp_server.outbound_credentials.sso_assertion_s rotate_sso_identity_assertions_master_key, ) from litellm.proxy._types import * -from litellm.proxy._types import LiteLLM_VerificationToken, hash_token +from litellm.proxy._types import Litellm_EntityType, LiteLLM_VerificationToken, hash_token from litellm.proxy.auth.auth_checks import ( _delete_cache_key_object, can_team_access_model, @@ -73,9 +74,7 @@ from litellm.proxy.common_utils.rbac_utils import check_org_admin_can_generate_k from litellm.proxy.common_utils.timezone_utils import get_budget_reset_time from litellm.proxy.common_utils.user_api_key_cache import UserApiKeyCache from litellm.proxy.hooks.key_management_event_hooks import KeyManagementEventHooks -from litellm.proxy.hooks.model_max_budget_limiter import ( - VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX, -) +from litellm.proxy.hooks.model_max_budget_limiter import build_model_max_budget_usage from litellm.proxy.management_endpoints.common_utils import ( _check_passthrough_routes_caller_permission, _is_user_org_admin_for_team, @@ -3511,62 +3510,17 @@ async def delete_key_fn( raise handle_exception_on_proxy(e) -async def _get_model_max_budget_current_spend( - api_key_hash: str, - model: str, - budget_config: BudgetConfig, - user_api_key_cache: UserApiKeyCache, -) -> float: - virtual_key_model_spend_cache_key = ( - f"{VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX}:{api_key_hash}:{model}:{budget_config.budget_duration}" - ) - current_spend: float | None = await user_api_key_cache.async_get_cache( - key=virtual_key_model_spend_cache_key, - ) - if current_spend is None: - model_without_prefix: Final = model.split("/")[-1] if "/" in model else model - virtual_key_model_spend_cache_key = ( - f"{VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX}:" - f"{api_key_hash}:{model_without_prefix}:{budget_config.budget_duration}" - ) - current_spend = await user_api_key_cache.async_get_cache( - key=virtual_key_model_spend_cache_key, - ) - try: - return float(current_spend or 0.0) - except (TypeError, ValueError): - return 0.0 - - async def _build_model_max_budget_usage( api_key_hash: str, model_max_budget: Mapping[str, Mapping[str, object]], - user_api_key_cache: UserApiKeyCache | None, + user_api_key_cache: DualCache | None, ) -> dict[str, dict[str, object]]: - if user_api_key_cache is None or not model_max_budget: - return {} - - result: Final[dict[str, dict[str, object]]] = {} - for model, budget_info in model_max_budget.items(): - try: - budget_config = BudgetConfig.model_validate(budget_info) - if budget_config.budget_duration is None: - continue - duration_in_seconds(budget_config.budget_duration) - except Exception: # noqa: BLE001 - continue - spend = await _get_model_max_budget_current_spend( - api_key_hash=api_key_hash, - model=model, - budget_config=budget_config, - user_api_key_cache=user_api_key_cache, - ) - result[model] = { - "current_spend": round(spend, 4), - "budget_limit": budget_config.max_budget, - "time_period": budget_config.budget_duration, - } - return result + return await build_model_max_budget_usage( + entity_type=Litellm_EntityType.KEY, + entity_id=api_key_hash, + model_max_budget=model_max_budget, + cache=user_api_key_cache, + ) @router.post( @@ -3596,7 +3550,10 @@ async def info_key_fn_v2( -d {"keys": ["sk-1", "sk-2", "sk-3"]} ``` """ - from litellm.proxy.proxy_server import prisma_client, user_api_key_cache + from litellm.proxy.proxy_server import ( + model_max_budget_limiter, + prisma_client, + ) try: if prisma_client is None: @@ -3648,7 +3605,7 @@ async def info_key_fn_v2( k_dict["model_max_budget_usage"] = await _build_model_max_budget_usage( api_key_hash=k_token_hash, model_max_budget=model_max_budget, - user_api_key_cache=user_api_key_cache, + user_api_key_cache=model_max_budget_limiter.dual_cache, ) filtered_key_info.append(k_dict) @@ -3707,7 +3664,10 @@ async def info_key_fn( -H "Authorization: Bearer sk-test-example-key-123" ``` """ - from litellm.proxy.proxy_server import prisma_client, user_api_key_cache + from litellm.proxy.proxy_server import ( + model_max_budget_limiter, + prisma_client, + ) try: if prisma_client is None: @@ -3760,7 +3720,7 @@ async def info_key_fn( key_info["model_max_budget_usage"] = await _build_model_max_budget_usage( api_key_hash=key_token_hash, model_max_budget=model_max_budget, - user_api_key_cache=user_api_key_cache, + user_api_key_cache=model_max_budget_limiter.dual_cache, ) # Attach object_permission if object_permission_id is set @@ -3953,6 +3913,10 @@ async def generate_key_helper_fn( } if teams is not None: user_data["teams"] = teams + if model_max_budget: + # Only when supplied: the SSO and default-key callers reach this with the + # empty default, and writing that would clear an existing user's budgets. + user_data["model_max_budget"] = model_max_budget_json key_data: Final = { "token": token, "key_alias": key_alias, diff --git a/litellm/proxy/pass_through_endpoints/pass_through_endpoints.py b/litellm/proxy/pass_through_endpoints/pass_through_endpoints.py index d45421489e7..1915a853983 100644 --- a/litellm/proxy/pass_through_endpoints/pass_through_endpoints.py +++ b/litellm/proxy/pass_through_endpoints/pass_through_endpoints.py @@ -64,6 +64,7 @@ from litellm.proxy._types import ( ProxyException, UserAPIKeyAuth, ) +from litellm.proxy.auth.auth_utils import request_dispatched_to_pass_through_endpoint from litellm.proxy.auth.user_api_key_auth import user_api_key_auth from litellm.proxy.common_request_processing import ( ProxyBaseLLMRequestProcessing, @@ -568,6 +569,22 @@ class HttpPassThroughEndpointHelpers(BasePassthroughUtils): _metadata["user_api_key"] = user_api_key_dict.api_key _metadata["litellm_parent_otel_span"] = user_api_key_dict.parent_otel_span _metadata["user_api_key_budget_reservation"] = user_api_key_dict.budget_reservation + # The per-model budget counters are keyed off these. get_sanitized_user_information_from_key + # returns StandardLoggingUserAPIKeyMetadata, which carries no budget field, so without this + # the post-call increment finds nothing and every passthrough request goes untracked and + # unenforced. Set after the client merge so a request body cannot supply its own budget. + # + # Only for the built-in provider routes. `get_model_from_request` returns + # None for a user-defined pass-through, deliberately: its body is forwarded + # verbatim, so `model` there names an UPSTREAM model rather than a + # LiteLLM-managed one. Enforcement is therefore skipped on those routes, and + # charging a counter anyway would track spend that nothing can refuse, and + # would attribute it to a budget the operator scoped to a LiteLLM model that + # merely shares the name. + if not request_dispatched_to_pass_through_endpoint(request): + _metadata["user_api_key_model_max_budget"] = user_api_key_dict.model_max_budget + _metadata["user_api_key_user_model_max_budget"] = user_api_key_dict.user_model_max_budget + _metadata["user_api_key_end_user_model_max_budget"] = user_api_key_dict.end_user_model_max_budget _metadata.update( LiteLLMProxyRequestSetup.get_sanitized_user_information_from_key(user_api_key_dict=user_api_key_dict) ) diff --git a/tests/proxy_unit_tests/test_unit_test_max_model_budget_limiter.py b/tests/proxy_unit_tests/test_unit_test_max_model_budget_limiter.py index 55459721906..fcdb3c9246c 100644 --- a/tests/proxy_unit_tests/test_unit_test_max_model_budget_limiter.py +++ b/tests/proxy_unit_tests/test_unit_test_max_model_budget_limiter.py @@ -2,16 +2,21 @@ import os import sys from unittest.mock import AsyncMock, patch -sys.path.insert( - 0, os.path.abspath("../..") -) # Adds the parent directory to the system-path +sys.path.insert(0, os.path.abspath("../..")) # Adds the parent directory to the system-path import pytest import litellm from litellm.caching.caching import DualCache +from datetime import datetime, timezone + +from litellm.litellm_core_utils.duration_parser import duration_in_seconds +from litellm.proxy._types import Litellm_EntityType from litellm.proxy.hooks.model_max_budget_limiter import ( + _budget_model_candidates, _PROXY_VirtualKeyModelMaxBudgetLimiter, + build_model_max_budget_usage, + resolve_model_budget, ) from litellm.proxy._types import UserAPIKeyAuth from litellm.types.utils import BudgetConfig as GenericBudgetInfo @@ -24,41 +29,95 @@ def budget_limiter(): return _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=dual_cache) -# Test _get_model_without_custom_llm_provider -def test_get_model_without_custom_llm_provider(budget_limiter): +# Test _budget_model_candidates +def test_budget_model_candidates(): # Test with custom provider - assert ( - budget_limiter._get_model_without_custom_llm_provider("openai/gpt-4") == "gpt-4" - ) + assert _budget_model_candidates("openai/gpt-4") == ("openai/gpt-4", "gpt-4") - # Test without custom provider - assert budget_limiter._get_model_without_custom_llm_provider("gpt-4") == "gpt-4" + # Test without custom provider: no duplicate candidate + assert _budget_model_candidates("gpt-4") == ("gpt-4",) -# Test _get_request_model_budget_config -def test_get_request_model_budget_config(budget_limiter): - internal_budget = { - "gpt-4": GenericBudgetInfo(budget_limit=100.0, time_period="1d"), - "claude-3": GenericBudgetInfo(budget_limit=50.0, time_period="1d"), +@pytest.mark.parametrize( + "model,expected", + [ + ( + "bedrock/anthropic.claude-opus-4-8", + ( + "bedrock/anthropic.claude-opus-4-8", + "anthropic.claude-opus-4-8", + "claude-opus-4-8", + ), + ), + ( + "us.anthropic.claude-opus-4-8", + ( + "us.anthropic.claude-opus-4-8", + "anthropic.claude-opus-4-8", + "claude-opus-4-8", + ), + ), + ( + "bedrock/converse/us.amazon.nova-pro-v1:0", + ( + "bedrock/converse/us.amazon.nova-pro-v1:0", + "us.amazon.nova-pro-v1:0", + "amazon.nova-pro-v1:0", + "nova-pro-v1:0", + ), + ), + ], +) +def test_budget_model_candidates_reach_the_bedrock_family_name(model, expected): + """ + Bedrock ids carry a dotted vendor segment ("anthropic.", "amazon.") on top of + the optional cross-region prefix, so a budget configured under the bare + family name would otherwise never match Bedrock traffic: no enforcement and + no spend tracking at all. + """ + assert _budget_model_candidates(model) == expected + + +@pytest.mark.parametrize( + "model", + [ + "azure/gpt-4.1", + "gpt-image-1.5", + "not-a-real-model.with.dots", + "ft:gpt-4o:acme::abc", + ], +) +def test_budget_model_candidates_never_split_a_non_bedrock_dotted_name(model): + """ + Most dotted model ids are versions, not Bedrock vendor prefixes. Splitting one + would offer a garbage candidate ("gpt-4.1" -> "1") that could collide with an + unrelated budget entry, so the split is gated on litellm pricing the model as + a Bedrock model. + """ + for candidate in _budget_model_candidates(model): + assert candidate in (model, model.split("/")[-1]) + + +# Test resolve_model_budget +def test_resolve_model_budget(): + model_max_budget = { + "gpt-4": {"budget_limit": 100.0, "time_period": "1d"}, + "claude-3": {"budget_limit": 50.0, "time_period": "1d"}, } # Test direct model match - config = budget_limiter._get_request_model_budget_config( - model="gpt-4", internal_model_max_budget=internal_budget - ) - assert config.max_budget == 100.0 + resolved = resolve_model_budget(model="gpt-4", model_max_budget=model_max_budget) + assert resolved.budget_model == "gpt-4" + assert resolved.budget_config.max_budget == 100.0 - # Test model with provider - config = budget_limiter._get_request_model_budget_config( - model="openai/gpt-4", internal_model_max_budget=internal_budget - ) - assert config.max_budget == 100.0 + # Test model with provider: the counter is keyed on the CONFIGURED name, + # not the request name, so every reader looks it up the same way. + resolved = resolve_model_budget(model="openai/gpt-4", model_max_budget=model_max_budget) + assert resolved.budget_model == "gpt-4" + assert resolved.budget_config.max_budget == 100.0 # Test non-existent model - config = budget_limiter._get_request_model_budget_config( - model="non-existent", internal_model_max_budget=internal_budget - ) - assert config is None + assert resolve_model_budget(model="non-existent", model_max_budget=model_max_budget) is None # Test is_key_within_model_budget @@ -72,47 +131,47 @@ async def test_is_key_within_model_budget(budget_limiter): ) # Test when model is within budget - with patch.object( - budget_limiter, "_get_virtual_key_spend_for_model", return_value=50.0 - ): - assert ( - await budget_limiter.is_key_within_model_budget(user_api_key, "gpt-4") - is True - ) + with patch.object(budget_limiter, "_get_spend_for_model_budget", return_value=50.0): + assert await budget_limiter.is_key_within_model_budget(user_api_key, "gpt-4") is True # Test when model exceeds budget - with patch.object( - budget_limiter, "_get_virtual_key_spend_for_model", return_value=150.0 - ): + with patch.object(budget_limiter, "_get_spend_for_model_budget", return_value=150.0): with pytest.raises(litellm.BudgetExceededError): await budget_limiter.is_key_within_model_budget(user_api_key, "gpt-4") # Test model not in budget config - assert ( - await budget_limiter.is_key_within_model_budget(user_api_key, "non-existent") - is True + assert await budget_limiter.is_key_within_model_budget(user_api_key, "non-existent") is True + + +# Test _get_spend_for_model_budget +@pytest.mark.asyncio +async def test_get_spend_for_model_budget_reads_the_configured_model_key( + budget_limiter, +): + from litellm.proxy.hooks.model_max_budget_limiter import ( + VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX, ) + model_max_budget = {"gpt-4": {"budget_limit": 100.0, "time_period": "1d"}} + # openai/gpt-4 resolves to the configured "gpt-4" entry, so the lookup must + # hit the same key async_log_success_event writes. + resolved = resolve_model_budget(model="openai/gpt-4", model_max_budget=model_max_budget) -# Test _get_virtual_key_spend_for_model -@pytest.mark.asyncio -async def test_get_virtual_key_spend_for_model(budget_limiter): - budget_config = GenericBudgetInfo(budget_limit=100.0, time_period="1d") + async def _spend(key): + return 50.0 if key == f"{VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX}:test-key:gpt-4:1d" else None - # Mock cache get - with patch.object(budget_limiter.dual_cache, "async_get_cache", return_value=50.0): - spend = await budget_limiter._get_virtual_key_spend_for_model( - user_api_key_hash="test-key", model="gpt-4", key_budget_config=budget_config - ) - assert spend == 50.0 - - # Test with provider prefix - spend = await budget_limiter._get_virtual_key_spend_for_model( - user_api_key_hash="test-key", + with patch.object(budget_limiter.dual_cache, "async_get_cache", side_effect=_spend) as mock_get: + spend = await budget_limiter._get_spend_for_model_budget( + entity_type=Litellm_EntityType.KEY, + entity_id="test-key", model="openai/gpt-4", - key_budget_config=budget_config, + resolved=resolved, ) assert spend == 50.0 + assert [call.kwargs["key"] for call in mock_get.call_args_list] == [ + f"{VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX}:test-key:gpt-4:1d", + f"{VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX}:test-key:openai/gpt-4:1d", + ] @pytest.mark.asyncio @@ -138,9 +197,7 @@ async def test_async_log_success_event_uses_per_model_budget_duration(budget_lim "metadata": {"user_api_key_hash": virtual_key}, }, "litellm_params": { - "metadata": { - "user_api_key_model_max_budget": user_api_key_model_max_budget - }, + "metadata": {"user_api_key_model_max_budget": user_api_key_model_max_budget}, }, } with patch.object( @@ -148,15 +205,11 @@ async def test_async_log_success_event_uses_per_model_budget_duration(budget_lim "_increment_spend_for_key", new_callable=AsyncMock, ) as mock_increment: - await budget_limiter.async_log_success_event( - kwargs, response_obj=None, start_time=None, end_time=None - ) + await budget_limiter.async_log_success_event(kwargs, response_obj=None, start_time=None, end_time=None) mock_increment.assert_awaited_once() call_kwargs = mock_increment.call_args.kwargs spend_key = call_kwargs["spend_key"] - assert spend_key == ( - f"{VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX}:{virtual_key}:{model}:{budget_duration}" - ) + assert spend_key == (f"{VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX}:{virtual_key}:{model}:{budget_duration}") assert call_kwargs["response_cost"] == 0.05 @@ -164,9 +217,7 @@ async def test_async_log_success_event_uses_per_model_budget_duration(budget_lim @pytest.mark.asyncio async def test_is_end_user_within_model_budget(budget_limiter): # Test when model is within budget - with patch.object( - budget_limiter, "_get_end_user_spend_for_model", return_value=50.0 - ): + with patch.object(budget_limiter, "_get_spend_for_model_budget", return_value=50.0): assert ( await budget_limiter.is_end_user_within_model_budget( "test-user", @@ -177,9 +228,7 @@ async def test_is_end_user_within_model_budget(budget_limiter): ) # Test when model exceeds budget - with patch.object( - budget_limiter, "_get_end_user_spend_for_model", return_value=150.0 - ): + with patch.object(budget_limiter, "_get_spend_for_model_budget", return_value=150.0): with pytest.raises(litellm.BudgetExceededError): await budget_limiter.is_end_user_within_model_budget( "test-user", @@ -198,25 +247,31 @@ async def test_is_end_user_within_model_budget(budget_limiter): ) -# Test _get_end_user_spend_for_model +# Test _get_spend_for_model_budget for the end-user scope @pytest.mark.asyncio -async def test_get_end_user_spend_for_model(budget_limiter): - budget_config = GenericBudgetInfo(budget_limit=100.0, time_period="1d") +async def test_get_spend_for_end_user_model_budget(budget_limiter): + from litellm.proxy.hooks.model_max_budget_limiter import ( + END_USER_SPEND_CACHE_KEY_PREFIX, + ) - # Mock cache get - with patch.object(budget_limiter.dual_cache, "async_get_cache", return_value=50.0): - spend = await budget_limiter._get_end_user_spend_for_model( - end_user_id="test-user", model="gpt-4", key_budget_config=budget_config - ) - assert spend == 50.0 + model_max_budget = {"gpt-4": {"budget_limit": 100.0, "time_period": "1d"}} + resolved = resolve_model_budget(model="openai/gpt-4", model_max_budget=model_max_budget) - # Test with provider prefix - spend = await budget_limiter._get_end_user_spend_for_model( - end_user_id="test-user", + async def _spend(key): + return 50.0 if key == f"{END_USER_SPEND_CACHE_KEY_PREFIX}:test-user:gpt-4:1d" else None + + with patch.object(budget_limiter.dual_cache, "async_get_cache", side_effect=_spend) as mock_get: + spend = await budget_limiter._get_spend_for_model_budget( + entity_type=Litellm_EntityType.END_USER, + entity_id="test-user", model="openai/gpt-4", - key_budget_config=budget_config, + resolved=resolved, ) assert spend == 50.0 + assert [call.kwargs["key"] for call in mock_get.call_args_list] == [ + f"{END_USER_SPEND_CACHE_KEY_PREFIX}:test-user:gpt-4:1d", + f"{END_USER_SPEND_CACHE_KEY_PREFIX}:test-user:openai/gpt-4:1d", + ] @pytest.mark.asyncio @@ -261,16 +316,12 @@ async def test_async_log_success_event_uses_model_group_for_cache_key(budget_lim "_increment_spend_for_key", new_callable=AsyncMock, ) as mock_increment: - await budget_limiter.async_log_success_event( - kwargs, response_obj=None, start_time=None, end_time=None - ) + await budget_limiter.async_log_success_event(kwargs, response_obj=None, start_time=None, end_time=None) mock_increment.assert_awaited_once() call_kwargs = mock_increment.call_args.kwargs spend_key = call_kwargs["spend_key"] # The cache key must use the model_group name, NOT the deployment name - assert spend_key == ( - f"{VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX}:{virtual_key}:{model_group}:{budget_duration}" - ) + assert spend_key == (f"{VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX}:{virtual_key}:{model_group}:{budget_duration}") assert call_kwargs["response_cost"] == 0.10 @@ -310,15 +361,11 @@ async def test_async_log_success_event_falls_back_to_model_when_no_model_group( "_increment_spend_for_key", new_callable=AsyncMock, ) as mock_increment: - await budget_limiter.async_log_success_event( - kwargs, response_obj=None, start_time=None, end_time=None - ) + await budget_limiter.async_log_success_event(kwargs, response_obj=None, start_time=None, end_time=None) mock_increment.assert_awaited_once() call_kwargs = mock_increment.call_args.kwargs spend_key = call_kwargs["spend_key"] - assert spend_key == ( - f"{VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX}:{virtual_key}:{model}:{budget_duration}" - ) + assert spend_key == (f"{VIRTUAL_KEY_SPEND_CACHE_KEY_PREFIX}:{virtual_key}:{model}:{budget_duration}") @pytest.mark.asyncio @@ -357,15 +404,11 @@ async def test_async_log_success_event_end_user_uses_model_group(budget_limiter) "_increment_spend_for_key", new_callable=AsyncMock, ) as mock_increment: - await budget_limiter.async_log_success_event( - kwargs, response_obj=None, start_time=None, end_time=None - ) + await budget_limiter.async_log_success_event(kwargs, response_obj=None, start_time=None, end_time=None) mock_increment.assert_awaited_once() call_kwargs = mock_increment.call_args.kwargs spend_key = call_kwargs["spend_key"] - assert spend_key == ( - f"{END_USER_SPEND_CACHE_KEY_PREFIX}:{end_user_id}:{model_group}:{budget_duration}" - ) + assert spend_key == (f"{END_USER_SPEND_CACHE_KEY_PREFIX}:{end_user_id}:{model_group}:{budget_duration}") @pytest.mark.asyncio @@ -393,9 +436,7 @@ async def test_async_log_success_event_uses_end_user_model_budget_duration( "metadata": {"user_api_key_end_user_id": end_user_id}, }, "litellm_params": { - "metadata": { - "user_api_key_end_user_model_max_budget": user_api_key_end_user_model_max_budget - }, + "metadata": {"user_api_key_end_user_model_max_budget": user_api_key_end_user_model_max_budget}, }, } with patch.object( @@ -403,15 +444,11 @@ async def test_async_log_success_event_uses_end_user_model_budget_duration( "_increment_spend_for_key", new_callable=AsyncMock, ) as mock_increment: - await budget_limiter.async_log_success_event( - kwargs, response_obj=None, start_time=None, end_time=None - ) + await budget_limiter.async_log_success_event(kwargs, response_obj=None, start_time=None, end_time=None) mock_increment.assert_awaited_once() call_kwargs = mock_increment.call_args.kwargs spend_key = call_kwargs["spend_key"] - assert spend_key == ( - f"{END_USER_SPEND_CACHE_KEY_PREFIX}:{end_user_id}:{model}:{budget_duration}" - ) + assert spend_key == (f"{END_USER_SPEND_CACHE_KEY_PREFIX}:{end_user_id}:{model}:{budget_duration}") assert call_kwargs["response_cost"] == 0.05 @@ -446,9 +483,7 @@ async def test_async_log_success_event_pushes_redis_increments_when_redis_config "_push_in_memory_increments_to_redis", new_callable=AsyncMock, ) as mock_push: - await limiter.async_log_success_event( - kwargs, response_obj=None, start_time=None, end_time=None - ) + await limiter.async_log_success_event(kwargs, response_obj=None, start_time=None, end_time=None) mock_push.assert_awaited_once() @@ -457,10 +492,7 @@ async def test_get_fallback_model_within_budget_returns_none_without_fallbacks( budget_limiter, ): user_api_key = UserAPIKeyAuth(token="test-key", budget_fallbacks={}) - assert ( - await budget_limiter.get_fallback_model_within_budget(user_api_key, "gpt-4") - is None - ) + assert await budget_limiter.get_fallback_model_within_budget(user_api_key, "gpt-4") is None @pytest.mark.asyncio @@ -472,12 +504,8 @@ async def test_get_fallback_model_within_budget_returns_first_within_budget( model_max_budget={"gpt-4o-mini": {"budget_limit": 100.0, "time_period": "1d"}}, budget_fallbacks={"gpt-4": ["gpt-4o-mini", "claude-haiku"]}, ) - with patch.object( - budget_limiter, "_get_virtual_key_spend_for_model", return_value=1.0 - ): - result = await budget_limiter.get_fallback_model_within_budget( - user_api_key, "gpt-4" - ) + with patch.object(budget_limiter, "_get_spend_for_model_budget", return_value=1.0): + result = await budget_limiter.get_fallback_model_within_budget(user_api_key, "gpt-4") assert result == "gpt-4o-mini" @@ -494,17 +522,15 @@ async def test_get_fallback_model_within_budget_skips_exhausted_fallback( budget_fallbacks={"gpt-4": ["gpt-4o-mini", "claude-haiku"]}, ) - async def _spend_for_model(user_api_key_hash, model, key_budget_config): - return 150.0 if model == "gpt-4o-mini" else 1.0 + async def _spend_for_model(entity_type, entity_id, model, resolved): + return 150.0 if resolved.budget_model == "gpt-4o-mini" else 1.0 with patch.object( budget_limiter, - "_get_virtual_key_spend_for_model", + "_get_spend_for_model_budget", side_effect=_spend_for_model, ): - result = await budget_limiter.get_fallback_model_within_budget( - user_api_key, "gpt-4" - ) + result = await budget_limiter.get_fallback_model_within_budget(user_api_key, "gpt-4") assert result == "claude-haiku" @@ -520,12 +546,8 @@ async def test_get_fallback_model_within_budget_returns_none_when_chain_exhauste }, budget_fallbacks={"gpt-4": ["gpt-4o-mini", "claude-haiku"]}, ) - with patch.object( - budget_limiter, "_get_virtual_key_spend_for_model", return_value=150.0 - ): - result = await budget_limiter.get_fallback_model_within_budget( - user_api_key, "gpt-4" - ) + with patch.object(budget_limiter, "_get_spend_for_model_budget", return_value=150.0): + result = await budget_limiter.get_fallback_model_within_budget(user_api_key, "gpt-4") assert result is None @@ -554,7 +576,761 @@ async def test_async_log_success_event_skips_redis_push_without_redis(budget_lim "_push_in_memory_increments_to_redis", new_callable=AsyncMock, ) as mock_push: - await budget_limiter.async_log_success_event( - kwargs, response_obj=None, start_time=None, end_time=None - ) + await budget_limiter.async_log_success_event(kwargs, response_obj=None, start_time=None, end_time=None) mock_push.assert_not_awaited() + + +def _success_kwargs( + *, + model_group, + deployment_model=None, + response_cost=0.5, + key_hash=None, + key_model_max_budget=None, + user_id=None, + user_model_max_budget=None, + end_user_id=None, + end_user_model_max_budget=None, +): + return { + "standard_logging_object": { + "response_cost": response_cost, + "model": deployment_model or model_group, + "model_group": model_group, + "end_user": end_user_id, + "metadata": { + "user_api_key_hash": key_hash, + "user_api_key_user_id": user_id, + "user_api_key_end_user_id": end_user_id, + }, + }, + "litellm_params": { + "metadata": { + "user_api_key_model_max_budget": key_model_max_budget, + "user_api_key_user_model_max_budget": user_model_max_budget, + "user_api_key_end_user_model_max_budget": end_user_model_max_budget, + }, + }, + } + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "request_model", + ["gpt-4", "openai/gpt-4"], + ids=["request_model_matches_budget_key", "request_model_carries_provider_prefix"], +) +async def test_logged_spend_is_visible_to_key_info_usage_and_enforcement(request_model): + """ + The counter written post-call, the counter enforcement reads and the counter + /key/info reports must be one and the same, including when the request model + is not byte-identical to the configured budget key. + + Regression: the increment used to be keyed on the REQUEST model while + /key/info only ever looked up the CONFIGURED model, so a key could be + actively blocked at 429 while reporting current_spend 0. + """ + dual_cache = DualCache() + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=dual_cache) + key_hash = "vk-hash" + model_max_budget = {"gpt-4": {"budget_limit": 1.0, "time_period": "1d"}} + + await limiter.async_log_success_event( + _success_kwargs( + model_group=request_model, + response_cost=0.75, + key_hash=key_hash, + key_model_max_budget=model_max_budget, + ), + response_obj=None, + start_time=None, + end_time=None, + ) + + usage = await build_model_max_budget_usage( + entity_type=Litellm_EntityType.KEY, + entity_id=key_hash, + model_max_budget=model_max_budget, + cache=dual_cache, + ) + assert usage == { + "gpt-4": { + "current_spend": 0.75, + "budget_limit": 1.0, + "time_period": "1d", + } + } + + user_api_key = UserAPIKeyAuth(token=key_hash, model_max_budget=model_max_budget) + # Still under the 1.0 limit. + assert await limiter.is_key_within_model_budget(user_api_key, request_model) is True + + await limiter.async_log_success_event( + _success_kwargs( + model_group=request_model, + response_cost=0.75, + key_hash=key_hash, + key_model_max_budget=model_max_budget, + ), + response_obj=None, + start_time=None, + end_time=None, + ) + with pytest.raises(litellm.BudgetExceededError): + await limiter.is_key_within_model_budget(user_api_key, request_model) + + usage_after = await build_model_max_budget_usage( + entity_type=Litellm_EntityType.KEY, + entity_id=key_hash, + model_max_budget=model_max_budget, + cache=dual_cache, + ) + assert usage_after["gpt-4"]["current_spend"] == 1.5 + + +@pytest.mark.asyncio +async def test_user_model_budget_is_tracked_and_enforced(): + """ + An internal user's own model_max_budget must be incremented post-call and + enforced, independently of any key-level budget. + """ + dual_cache = DualCache() + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=dual_cache) + user_id = "user-1" + user_model_max_budget = {"gpt-4": {"budget_limit": 1.0, "time_period": "1mo"}} + + assert ( + await limiter.is_user_within_model_budget( + user_id=user_id, + user_model_max_budget=user_model_max_budget, + model="openai/gpt-4", + ) + is True + ) + + await limiter.async_log_success_event( + _success_kwargs( + model_group="openai/gpt-4", + response_cost=1.5, + user_id=user_id, + user_model_max_budget=user_model_max_budget, + ), + response_obj=None, + start_time=None, + end_time=None, + ) + + assert await build_model_max_budget_usage( + entity_type=Litellm_EntityType.USER, + entity_id=user_id, + model_max_budget=user_model_max_budget, + cache=dual_cache, + ) == {"gpt-4": {"current_spend": 1.5, "budget_limit": 1.0, "time_period": "1mo"}} + + with pytest.raises(litellm.BudgetExceededError) as exc: + await limiter.is_user_within_model_budget( + user_id=user_id, + user_model_max_budget=user_model_max_budget, + model="openai/gpt-4", + ) + assert exc.value.entity_type == Litellm_EntityType.USER.value + + +@pytest.mark.asyncio +async def test_user_model_budget_counter_is_separate_from_the_key_counter(): + """ + A key budget and a user budget over the same model are two independent + counters, so one request must charge each exactly once. + """ + dual_cache = DualCache() + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=dual_cache) + model_max_budget = {"gpt-4": {"budget_limit": 10.0, "time_period": "1d"}} + + await limiter.async_log_success_event( + _success_kwargs( + model_group="gpt-4", + response_cost=2.0, + key_hash="vk-hash", + key_model_max_budget=model_max_budget, + user_id="user-1", + user_model_max_budget=model_max_budget, + ), + response_obj=None, + start_time=None, + end_time=None, + ) + + assert await dual_cache.async_get_cache(key="virtual_key_spend:vk-hash:gpt-4:1d") == 2.0 + assert await dual_cache.async_get_cache(key="user_model_spend:user-1:gpt-4:1d") == 2.0 + + +@pytest.mark.asyncio +async def test_two_models_on_one_key_do_not_share_a_budget_window(): + """ + A key budgeting two models over different periods must own one window start + per model: a shared start lets the shorter period restart the longer one. + """ + dual_cache = DualCache() + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=dual_cache) + model_max_budget = { + "gpt-4": {"budget_limit": 10.0, "time_period": "1d"}, + "claude-3": {"budget_limit": 10.0, "time_period": "30d"}, + } + + start_time_keys = [] + with patch.object(limiter, "_increment_spend_for_key", new_callable=AsyncMock) as mock_increment: + for model in ("gpt-4", "claude-3"): + await limiter.async_log_success_event( + _success_kwargs( + model_group=model, + key_hash="vk-hash", + key_model_max_budget=model_max_budget, + ), + response_obj=None, + start_time=None, + end_time=None, + ) + start_time_keys = [call.kwargs["start_time_key"] for call in mock_increment.call_args_list] + + assert start_time_keys == [ + "virtual_key_budget_start_time:vk-hash:gpt-4:1d", + "virtual_key_budget_start_time:vk-hash:claude-3:30d", + ] + assert len(set(start_time_keys)) == 2 + + +@pytest.mark.asyncio +async def test_no_increment_when_no_scope_budgets_the_model(): + dual_cache = DualCache() + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=dual_cache) + with patch.object(limiter, "_increment_spend_for_key", new_callable=AsyncMock) as mock_increment: + await limiter.async_log_success_event( + _success_kwargs( + model_group="gpt-4", + key_hash="vk-hash", + key_model_max_budget={"claude-3": {"budget_limit": 1.0, "time_period": "1d"}}, + user_id="user-1", + user_model_max_budget={}, + ), + response_obj=None, + start_time=None, + end_time=None, + ) + mock_increment.assert_not_awaited() + + +@pytest.mark.asyncio +async def test_build_model_max_budget_usage_skips_unusable_entries(): + """A malformed or period-less entry must be omitted, not crash the report.""" + dual_cache = DualCache() + await dual_cache.async_set_cache(key="virtual_key_spend:vk:gpt-4:1d", value=3.0) + + usage = await build_model_max_budget_usage( + entity_type=Litellm_EntityType.KEY, + entity_id="vk", + model_max_budget={ + "gpt-4": {"budget_limit": 10.0, "time_period": "1d"}, + "no-period": {"budget_limit": 10.0}, + "bad-period": {"budget_limit": 10.0, "time_period": "not-a-duration"}, + }, + cache=dual_cache, + ) + assert usage == {"gpt-4": {"current_spend": 3.0, "budget_limit": 10.0, "time_period": "1d"}} + + +@pytest.mark.asyncio +async def test_bedrock_traffic_charges_the_bare_family_name_budget(): + """ + The reported case: a budget configured as "claude-opus-4-8" with traffic on + "bedrock/anthropic.claude-opus-4-8". Before the fix nothing matched, so spend + was never tracked and the budget was never enforced no matter how far over it + the key went. + """ + dual_cache = DualCache() + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=dual_cache) + key_hash = "vk-hash" + model_max_budget = {"claude-opus-4-8": {"budget_limit": 1.0, "time_period": "18h"}} + user_api_key = UserAPIKeyAuth(token=key_hash, model_max_budget=model_max_budget) + + await limiter.async_log_success_event( + _success_kwargs( + model_group="bedrock/anthropic.claude-opus-4-8", + response_cost=1.5, + key_hash=key_hash, + key_model_max_budget=model_max_budget, + ), + response_obj=None, + start_time=None, + end_time=None, + ) + + assert await build_model_max_budget_usage( + entity_type=Litellm_EntityType.KEY, + entity_id=key_hash, + model_max_budget=model_max_budget, + cache=dual_cache, + ) == { + "claude-opus-4-8": { + "current_spend": 1.5, + "budget_limit": 1.0, + "time_period": "18h", + } + } + + with pytest.raises(litellm.BudgetExceededError): + await limiter.is_key_within_model_budget(user_api_key, "bedrock/anthropic.claude-opus-4-8") + + +@pytest.mark.asyncio +async def test_user_model_budget_window_resets_when_the_period_elapses(): + """ + A monthly user budget must start a fresh window once the period elapses, + and the window start must be scoped to that one budget model so a second + model on a shorter period cannot drag it forward. + """ + from litellm.proxy.hooks.model_max_budget_limiter import ( + model_budget_spend_cache_key, + model_budget_start_time_cache_key, + ) + + dual_cache = DualCache() + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=dual_cache) + user_id = "user-1" + user_model_max_budget = {"gpt-4": {"budget_limit": 1.0, "time_period": "1mo"}} + spend_key = model_budget_spend_cache_key( + entity_type=Litellm_EntityType.USER, + entity_id=user_id, + budget_model="gpt-4", + budget_duration="1mo", + ) + start_time_key = model_budget_start_time_cache_key( + entity_type=Litellm_EntityType.USER, + entity_id=user_id, + budget_model="gpt-4", + budget_duration="1mo", + ) + + kwargs = _success_kwargs( + model_group="gpt-4", + response_cost=1.5, + user_id=user_id, + user_model_max_budget=user_model_max_budget, + ) + await limiter.async_log_success_event(kwargs, response_obj=None, start_time=None, end_time=None) + assert await dual_cache.async_get_cache(key=spend_key) == 1.5 + with pytest.raises(litellm.BudgetExceededError): + await limiter.is_user_within_model_budget( + user_id=user_id, + user_model_max_budget=user_model_max_budget, + model="gpt-4", + ) + + # Age the window past its period. The next charge opens a new window rather + # than adding to the exhausted one. + elapsed = duration_in_seconds("1mo") + 60 + await dual_cache.async_set_cache( + key=start_time_key, + value=datetime.now(timezone.utc).timestamp() - elapsed, + ttl=elapsed, + ) + + await limiter.async_log_success_event(kwargs, response_obj=None, start_time=None, end_time=None) + assert await dual_cache.async_get_cache(key=spend_key) == 1.5 + assert await build_model_max_budget_usage( + entity_type=Litellm_EntityType.USER, + entity_id=user_id, + model_max_budget=user_model_max_budget, + cache=dual_cache, + ) == {"gpt-4": {"current_spend": 1.5, "budget_limit": 1.0, "time_period": "1mo"}} + + +@pytest.mark.asyncio +async def test_a_zero_dollar_cap_blocks_the_model(): + """ + 0 is the operator saying "nobody may spend anything on this model", which is + the strictest cap expressible, not the absence of one. Skipping it on + falsiness turned the strictest setting into no setting at all, so the model + stayed wide open. The dashboard editor can produce this value, so it has to + mean something. + """ + dual_cache = DualCache() + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=dual_cache) + key = UserAPIKeyAuth( + token="hash-zero", + model_max_budget={"gpt-4": {"budget_limit": 0, "time_period": "1d"}}, + ) + + with pytest.raises(litellm.BudgetExceededError) as exc: + await limiter.is_key_within_model_budget(user_api_key_dict=key, model="gpt-4") + assert exc.value.max_budget == 0 + + +@pytest.mark.asyncio +async def test_a_zero_dollar_cap_is_reported_as_a_cap_not_as_absent(): + """The usage endpoints must show the 0 too, or an operator cannot see the block they configured.""" + assert await build_model_max_budget_usage( + entity_type=Litellm_EntityType.KEY, + entity_id="hash-zero", + model_max_budget={"gpt-4": {"budget_limit": 0, "time_period": "1d"}}, + cache=DualCache(), + ) == {"gpt-4": {"current_spend": 0.0, "budget_limit": 0.0, "time_period": "1d"}} + + +@pytest.mark.asyncio +async def test_spend_exactly_at_the_cap_is_refused(): + """ + Spending the whole budget exhausts it. `>` let a caller sit exactly on the + limit and keep going, and every sibling budget check in the codebase + (RouterBudgetLimiting, the key and team budget checks) uses `>=`. + """ + dual_cache = DualCache() + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=dual_cache) + budget = {"gpt-4": {"budget_limit": 2.0, "time_period": "1d"}} + key = UserAPIKeyAuth(token="hash-exact", model_max_budget=budget) + + await limiter.async_log_success_event( + _success_kwargs(model_group="gpt-4", response_cost=2.0, key_hash="hash-exact", key_model_max_budget=budget), + response_obj=None, + start_time=None, + end_time=None, + ) + + with pytest.raises(litellm.BudgetExceededError): + await limiter.is_key_within_model_budget(user_api_key_dict=key, model="gpt-4") + + +@pytest.mark.asyncio +async def test_usage_report_reads_every_counter_in_one_batched_lookup(): + """ + model_max_budget is caller-supplied and unbounded in size, so one cache + coroutine per configured model let a large map fan out into an unbounded + number of concurrent lookups on an endpoint anyone holding the key can call. + One batched read keeps it to a single round trip whatever the map's size. + """ + dual_cache = DualCache() + budget = {f"model-{i}": {"budget_limit": 1.0, "time_period": "1d"} for i in range(50)} + + with ( + patch.object(dual_cache, "async_batch_get_cache", new=AsyncMock(return_value=[None] * 50)) as batched, + patch.object(dual_cache, "async_get_cache", new=AsyncMock()) as single, + ): + usage = await build_model_max_budget_usage( + entity_type=Litellm_EntityType.KEY, + entity_id="hash-many", + model_max_budget=budget, + cache=dual_cache, + ) + + assert batched.await_count == 1 + assert len(batched.await_args.kwargs["keys"]) == 50 + assert single.await_count == 0 + assert len(usage) == 50 + + +@pytest.mark.asyncio +async def test_usage_report_survives_a_batch_lookup_that_returns_nothing(): + """ + async_batch_get_cache swallows its own failures and returns None. Zipping + that against the budgets would raise and take the whole /key/info response + with it, so an unusable result has to read as a miss instead. + """ + dual_cache = DualCache() + with patch.object(dual_cache, "async_batch_get_cache", new=AsyncMock(return_value=None)): + assert await build_model_max_budget_usage( + entity_type=Litellm_EntityType.KEY, + entity_id="hash-none", + model_max_budget={"gpt-4": {"budget_limit": 1.0, "time_period": "1d"}}, + cache=dual_cache, + ) == {"gpt-4": {"current_spend": 0.0, "budget_limit": 1.0, "time_period": "1d"}} + + +@pytest.mark.asyncio +async def test_one_malformed_scope_does_not_abort_the_other_scopes(): + """ + Every scope is resolved before any of them is incremented, so a single + unusable entry used to raise out of resolution and leave the key counter + unwritten too. The key's budget is well formed here and must still be + charged despite the user's entry being garbage. + """ + dual_cache = DualCache() + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=dual_cache) + key_budget = {"gpt-4": {"budget_limit": 10.0, "time_period": "1d"}} + + await limiter.async_log_success_event( + _success_kwargs( + model_group="gpt-4", + response_cost=0.25, + key_hash="hash-mixed", + key_model_max_budget=key_budget, + user_id="user-mixed", + user_model_max_budget={"gpt-4": {"budget_limit": "not-a-number", "time_period": "1d"}}, + ), + response_obj=None, + start_time=None, + end_time=None, + ) + + assert await build_model_max_budget_usage( + entity_type=Litellm_EntityType.KEY, + entity_id="hash-mixed", + model_max_budget=key_budget, + cache=dual_cache, + ) == {"gpt-4": {"current_spend": 0.25, "budget_limit": 10.0, "time_period": "1d"}} + + +@pytest.mark.asyncio +async def test_an_unusable_budget_entry_is_not_enforced_instead_of_raising(): + """ + A config typo must not turn every request for that model into a 500. It + cannot be keyed, so it cannot be enforced; the write path rejects these, so + reaching here means config.yaml or a direct DB edit. + """ + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=DualCache()) + key = UserAPIKeyAuth( + token="hash-malformed", + model_max_budget={"gpt-4": {"budget_limit": "not-a-number", "time_period": "1d"}}, + ) + + assert await limiter.is_key_within_model_budget(user_api_key_dict=key, model="gpt-4") is True + + +def test_resolve_model_budget_returns_none_for_an_unusable_entry(): + assert ( + resolve_model_budget( + model="gpt-4", + model_max_budget={"gpt-4": {"budget_limit": "not-a-number", "time_period": "1d"}}, + ) + is None + ) + + +def test_a_malformed_specific_entry_does_not_hide_a_usable_family_budget(): + """ + The candidate chain is most-specific-first and already falls through an entry + that is ABSENT. An entry that will not parse is indistinguishable from absent + as far as enforcement goes, so it has to fall through too: otherwise one bad + provider-prefixed entry silently disables the valid bare-family budget sitting + next to it, and the model goes uncapped. + """ + resolved = resolve_model_budget( + model="openai/gpt-4", + model_max_budget={ + "openai/gpt-4": {"budget_limit": "not-a-number", "time_period": "1d"}, + "gpt-4": {"budget_limit": 7.0, "time_period": "1d"}, + }, + ) + + assert resolved is not None + assert resolved.budget_model == "gpt-4" + assert resolved.budget_config.max_budget == 7.0 + + +@pytest.mark.asyncio +async def test_a_malformed_specific_entry_still_enforces_the_family_budget(): + """The fall-through has to reach enforcement, not just resolution.""" + dual_cache = DualCache() + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=dual_cache) + budget = { + "openai/gpt-4": {"budget_limit": "not-a-number", "time_period": "1d"}, + "gpt-4": {"budget_limit": 1.0, "time_period": "1d"}, + } + key = UserAPIKeyAuth(token="hash-fallthrough", model_max_budget=budget) + + await limiter.async_log_success_event( + _success_kwargs( + model_group="openai/gpt-4", + response_cost=2.0, + key_hash="hash-fallthrough", + key_model_max_budget=budget, + ), + response_obj=None, + start_time=None, + end_time=None, + ) + + with pytest.raises(litellm.BudgetExceededError): + await limiter.is_key_within_model_budget(user_api_key_dict=key, model="openai/gpt-4") + + +def test_documented_budget_spelling_survives_model_validate(): + """ + `budget_limit` / `time_period` are the spelling the docs, the CRUD endpoints + and the dashboard editor all use, and BudgetConfig maps them onto + `max_budget` / `budget_duration` inside its `__init__`. + + Pydantic v2 normally bypasses a custom `__init__` in `model_validate`, and + this code path validates rather than constructing. It works today, but that + is a property of the installed Pydantic rather than of anything in this + repository, so an upgrade could silently stop applying the mapping and + quietly disable every budget written in the documented spelling. Pinned here + so that becomes a red test instead of an outage. + """ + from litellm.types.utils import BudgetConfig + + validated = BudgetConfig.model_validate({"budget_limit": 5, "time_period": "1d"}) + assert validated.max_budget == 5.0 + assert validated.budget_duration == "1d" + + # Control: an unrecognised key must NOT populate max_budget, or the assertion + # above would also pass against a model that accepted anything at all. + ignored = BudgetConfig.model_validate({"bogus_limit": 5, "time_period": "1d"}) + assert ignored.max_budget is None + + +def test_resolution_accepts_both_documented_spellings(): + """The resolver is what enforcement, tracking and reporting all go through.""" + for budget in ( + {"gpt-4": {"budget_limit": 5, "time_period": "1d"}}, + {"gpt-4": {"max_budget": 5, "budget_duration": "1d"}}, + ): + resolved = resolve_model_budget(model="gpt-4", model_max_budget=budget) + assert resolved is not None, f"{budget} resolved to nothing" + assert resolved.budget_config.max_budget == 5.0 + assert resolved.budget_config.budget_duration == "1d" + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "entity_type, prefix", + [ + (Litellm_EntityType.KEY, "virtual_key_spend"), + (Litellm_EntityType.END_USER, "end_user_model_spend"), + ], +) +async def test_a_pre_upgrade_counter_keyed_on_the_request_model_still_enforces(entity_type, prefix): + """An upgrading proxy must not hand out a second allowance for the window it is already in. + + Before the counter key moved to the configured budget model, spend for a + request on `openai/gpt-4` against a budget configured as `gpt-4` was both + written to and enforced on `{prefix}:{id}:openai/gpt-4:1d`. Reading only the + configured-model key finds that counter empty and admits another full budget + until the window expires. + """ + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=DualCache()) + model_max_budget = {"gpt-4": {"budget_limit": 10.0, "time_period": "1d"}} + await limiter.dual_cache.async_set_cache(key=f"{prefix}:entity-1:openai/gpt-4:1d", value=25.0, ttl=86400) + + with pytest.raises(litellm.BudgetExceededError) as exc_info: + if entity_type == Litellm_EntityType.KEY: + await limiter.is_key_within_model_budget( + user_api_key_dict=UserAPIKeyAuth(token="entity-1", model_max_budget=model_max_budget), + model="openai/gpt-4", + ) + else: + await limiter.is_end_user_within_model_budget( + end_user_id="entity-1", + end_user_model_max_budget=model_max_budget, + model="openai/gpt-4", + ) + assert exc_info.value.current_cost == 25.0 + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "legacy_spend, current_spend, expect_blocked", + [(6.0, 5.0, True), (2.0, 3.0, False)], +) +async def test_the_pre_upgrade_and_post_upgrade_counters_add_up_over_one_window( + legacy_spend, current_spend, expect_blocked +): + """The two counters hold disjoint halves of one window, so the window's spend is their sum. + + Nothing writes the request-model spelling once this version is running, so + the legacy counter is frozen at whatever the previous version charged and + the configured-model counter carries everything since. Either one alone + under-reports the window: 6 + 5 is over a cap of 10 that neither half + reaches on its own. + """ + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=DualCache()) + model_max_budget = {"gpt-4": {"budget_limit": 10.0, "time_period": "1d"}} + await limiter.dual_cache.async_set_cache( + key="virtual_key_spend:entity-1:openai/gpt-4:1d", value=legacy_spend, ttl=86400 + ) + await limiter.dual_cache.async_set_cache(key="virtual_key_spend:entity-1:gpt-4:1d", value=current_spend, ttl=86400) + + async def enforce(): + return await limiter.is_key_within_model_budget( + user_api_key_dict=UserAPIKeyAuth(token="entity-1", model_max_budget=model_max_budget), + model="openai/gpt-4", + ) + + if expect_blocked: + with pytest.raises(litellm.BudgetExceededError) as exc_info: + await enforce() + assert exc_info.value.current_cost == legacy_spend + current_spend + else: + assert await enforce() is True + + +@pytest.mark.asyncio +async def test_the_configured_model_counter_is_never_counted_twice(): + """When the request names the budget exactly there is no legacy counter, only the one key. + + Both keys are `virtual_key_spend:entity-1:gpt-4:1d` here, so a lookup that + added them without noticing would charge 12 against a cap of 10 and refuse a + key that has spent 6. + """ + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=DualCache()) + await limiter.dual_cache.async_set_cache(key="virtual_key_spend:entity-1:gpt-4:1d", value=6.0, ttl=86400) + + assert ( + await limiter.is_key_within_model_budget( + user_api_key_dict=UserAPIKeyAuth( + token="entity-1", + model_max_budget={"gpt-4": {"budget_limit": 10.0, "time_period": "1d"}}, + ), + model="gpt-4", + ) + is True + ) + + +@pytest.mark.asyncio +async def test_the_pre_upgrade_counter_is_no_longer_read_a_window_after_start_up(monkeypatch): + """The carry is bounded, so it cannot become a permanent second lookup on every request. + + A counter written by the previous version belongs to a window that was + already open when this process replaced it, so once a full window has passed + since start-up there is nothing left for the lookup to find. + """ + import litellm.proxy.hooks.model_max_budget_limiter as limiter_module + + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=DualCache()) + model_max_budget = {"gpt-4": {"budget_limit": 10.0, "time_period": "1d"}} + await limiter.dual_cache.async_set_cache(key="virtual_key_spend:entity-1:openai/gpt-4:1d", value=25.0, ttl=86400) + user_api_key = UserAPIKeyAuth(token="entity-1", model_max_budget=model_max_budget) + + # Control: within the first window since start-up the same counter blocks, + # so the assertion below cannot pass against a lookup that never worked. + with pytest.raises(litellm.BudgetExceededError): + await limiter.is_key_within_model_budget(user_api_key_dict=user_api_key, model="openai/gpt-4") + + monkeypatch.setattr(limiter_module, "_PROCESS_STARTED_AT", limiter_module.time.monotonic() - 86401) + assert await limiter.is_key_within_model_budget(user_api_key_dict=user_api_key, model="openai/gpt-4") is True + + +@pytest.mark.asyncio +async def test_the_user_scope_has_no_pre_upgrade_counter_to_carry(): + """The user scope is introduced by this change, so a request-model key under it is not one of ours. + + Reading one would invent a counter no previous version ever wrote, which is + the opposite of preserving one. + """ + limiter = _PROXY_VirtualKeyModelMaxBudgetLimiter(dual_cache=DualCache()) + model_max_budget = {"gpt-4": {"budget_limit": 10.0, "time_period": "1d"}} + await limiter.dual_cache.async_set_cache(key="user_model_spend:u1:openai/gpt-4:1d", value=25.0, ttl=86400) + + assert ( + await limiter.is_user_within_model_budget( + user_id="u1", user_model_max_budget=model_max_budget, model="openai/gpt-4" + ) + is True + ) + + # Control: the same overspend under the key this scope does own must block, + # or the assertion above would pass against a scope that enforces nothing. + await limiter.dual_cache.async_set_cache(key="user_model_spend:u1:gpt-4:1d", value=25.0, ttl=86400) + with pytest.raises(litellm.BudgetExceededError): + await limiter.is_user_within_model_budget( + user_id="u1", user_model_max_budget=model_max_budget, model="openai/gpt-4" + ) diff --git a/tests/proxy_unit_tests/test_user_api_key_auth.py b/tests/proxy_unit_tests/test_user_api_key_auth.py index 58dbe3ad370..e9566254dbc 100644 --- a/tests/proxy_unit_tests/test_user_api_key_auth.py +++ b/tests/proxy_unit_tests/test_user_api_key_auth.py @@ -7,9 +7,7 @@ import sys import litellm.proxy import litellm.proxy.proxy_server -sys.path.insert( - 0, os.path.abspath("../..") -) # Adds the parent directory to the system path +sys.path.insert(0, os.path.abspath("../..")) # Adds the parent directory to the system path from typing import Dict, List, Optional from unittest.mock import MagicMock, patch, AsyncMock @@ -50,9 +48,7 @@ class Request: ), # Request with no client IP should not be allowed ], ) -def test_check_valid_ip( - allowed_ips: Optional[List[str]], client_ip: Optional[str], expected_result: bool -): +def test_check_valid_ip(allowed_ips: Optional[List[str]], client_ip: Optional[str], expected_result: bool): from litellm.proxy.auth.auth_utils import _check_valid_ip request = Request(client_ip) @@ -121,9 +117,7 @@ async def test_check_blocked_team(): last_refreshed_at=time.time(), ) await asyncio.sleep(1) - team_obj = LiteLLM_TeamTableCachedObj( - team_id=_team_id, blocked=False, last_refreshed_at=time.time() - ) + team_obj = LiteLLM_TeamTableCachedObj(team_id=_team_id, blocked=False, last_refreshed_at=time.time()) hashed_token = hash_token(user_key) print(f"STORING TOKEN UNDER KEY={hashed_token}") user_api_key_cache.set_cache(key=hashed_token, value=valid_token) @@ -173,9 +167,7 @@ async def test_team_object_has_object_permission_id(): request = Request(scope={"type": "http"}) request._url = URL(url="/chat/completions") - with patch( - "litellm.proxy.auth.user_api_key_auth.common_checks", new_callable=AsyncMock - ) as mock_common_checks: + with patch("litellm.proxy.auth.user_api_key_auth.common_checks", new_callable=AsyncMock) as mock_common_checks: mock_common_checks.return_value = True await user_api_key_auth(request=request, api_key="Bearer " + user_key) @@ -200,9 +192,7 @@ async def test_returned_user_api_key_auth(user_role, expected_role): from datetime import datetime new_obj = await _return_user_api_key_auth_obj( - user_obj=LiteLLM_UserTable( - user_role=user_role, user_id="", max_budget=None, user_email="" - ), + user_obj=LiteLLM_UserTable(user_role=user_role, user_id="", max_budget=None, user_email=""), api_key="hello-world", parent_otel_span=None, valid_token_dict={}, @@ -258,9 +248,7 @@ async def test_aaauser_personal_budgets(key_ownership): spend=20, ) - user_obj = LiteLLM_UserTable( - user_id=_user_id, spend=11, max_budget=10, user_email="" - ) + user_obj = LiteLLM_UserTable(user_id=_user_id, spend=11, max_budget=10, user_email="") user_api_key_cache.set_cache(key=hash_token(user_key), value=valid_token) user_api_key_cache.set_cache(key="{}".format(_user_id), value=user_obj) @@ -273,10 +261,7 @@ async def test_aaauser_personal_budgets(key_ownership): test_user_cache = getattr(litellm.proxy.proxy_server, "user_api_key_cache") - assert ( - test_user_cache.get_cache(key=hash_token(user_key), model_type=UserAPIKeyAuth) - == valid_token - ) + assert test_user_cache.get_cache(key=hash_token(user_key), model_type=UserAPIKeyAuth) == valid_token if key_ownership == "user_key": with pytest.raises(ProxyException) as exc_info: @@ -311,9 +296,7 @@ async def test_user_api_key_auth_fails_with_prohibited_params(prohibited_param): request.body = return_body try: - response = await user_api_key_auth( - request=request, api_key="Bearer " + user_key - ) + response = await user_api_key_auth(request=request, api_key="Bearer " + user_key) except Exception as e: print("error str=", str(e)) error_message = str(e.message) @@ -519,9 +502,7 @@ def _assert_api_key_from_custom_header(headers, custom_header_name, expected_api verbose_proxy_logger.setLevel(logging.DEBUG) request = MagicMock(spec=Request) request.headers = headers - api_key = get_api_key_from_custom_header( - request=request, custom_litellm_key_header_name=custom_header_name - ) + api_key = get_api_key_from_custom_header(request=request, custom_litellm_key_header_name=custom_header_name) assert api_key == expected_api_key @@ -572,9 +553,7 @@ from litellm.proxy._types import LitellmUserRoles (LitellmUserRoles.TEAM, "1234", "1234", True), ], ) -def test_allowed_route_inside_route( - user_role, auth_user_id, requested_user_id, expected_result -): +def test_allowed_route_inside_route(user_role, auth_user_id, requested_user_id, expected_result): from litellm.proxy.auth.auth_checks import allowed_route_check_inside_route from litellm.proxy._types import UserAPIKeyAuth, LitellmUserRoles @@ -715,9 +694,7 @@ async def test_soft_budget_alert(): try: # Call user_api_key_auth - response = await user_api_key_auth( - request=request, api_key="Bearer " + user_key - ) + response = await user_api_key_auth(request=request, api_key="Bearer " + user_key) # Assert the request was allowed (no exception raised) assert response is not None @@ -883,9 +860,7 @@ async def test_user_api_key_auth_websocket(): mock_websocket.url = URL(url="/ws") # Mock the return value of `user_api_key_auth` when it's called within the `user_api_key_auth_websocket` function - with patch( - "litellm.proxy.auth.user_api_key_auth.user_api_key_auth", autospec=True - ) as mock_user_api_key_auth: + with patch("litellm.proxy.auth.user_api_key_auth.user_api_key_auth", autospec=True) as mock_user_api_key_auth: # Make the call to the WebSocket function await user_api_key_auth_websocket(mock_websocket) @@ -896,17 +871,11 @@ async def test_user_api_key_auth_websocket(): request_arg = mock_user_api_key_auth.call_args.kwargs["request"] # Verify that the request has headers set - assert hasattr( - request_arg, "headers" - ), "Request object should have headers attribute" - assert ( - "authorization" in request_arg.headers - ), "Request headers should contain authorization" + assert hasattr(request_arg, "headers"), "Request object should have headers attribute" + assert "authorization" in request_arg.headers, "Request headers should contain authorization" assert request_arg.headers["authorization"] == "Bearer some_api_key" - assert ( - mock_user_api_key_auth.call_args.kwargs["api_key"] == "Bearer some_api_key" - ) + assert mock_user_api_key_auth.call_args.kwargs["api_key"] == "Bearer some_api_key" @pytest.mark.asyncio @@ -929,9 +898,7 @@ async def test_user_api_key_auth_websocket_carries_asgi_path(): } mock_websocket.url = URL(url="/v1/realtime") - with patch( - "litellm.proxy.auth.user_api_key_auth.user_api_key_auth", autospec=True - ) as mock_user_api_key_auth: + with patch("litellm.proxy.auth.user_api_key_auth.user_api_key_auth", autospec=True) as mock_user_api_key_auth: await user_api_key_auth_websocket(mock_websocket) request_arg = mock_user_api_key_auth.call_args.kwargs["request"] @@ -1127,9 +1094,7 @@ async def test_jwt_non_admin_team_route_access(monkeypatch): ) request._url = URL(url="/team/new") - monkeypatch.setattr( - litellm.proxy.proxy_server, "general_settings", {"enable_jwt_auth": True} - ) + monkeypatch.setattr(litellm.proxy.proxy_server, "general_settings", {"enable_jwt_auth": True}) # Initialize jwt_handler with a default LiteLLM_JWTAuth so that the # virtual_key_claim_field check in user_api_key_auth doesn't fail with @@ -1158,9 +1123,7 @@ async def test_jwt_non_admin_team_route_access(monkeypatch): ): try: await user_api_key_auth(request=request, api_key="Bearer fake.jwt.token") - pytest.fail( - "Expected this call to fail. Non-admin user should not access team routes." - ) + pytest.fail("Expected this call to fail. Non-admin user should not access team routes.") except ProxyException as e: print("e", e) assert "Only proxy admin can be used to generate" in str(e.message) @@ -1220,9 +1183,7 @@ async def test_user_api_key_from_query_param(): from litellm.proxy.proxy_server import hash_token, user_api_key_cache user_key = "sk-query-1234" - user_api_key_cache.set_cache( - key=hash_token(user_key), value=UserAPIKeyAuth(token=hash_token(user_key)) - ) + user_api_key_cache.set_cache(key=hash_token(user_key), value=UserAPIKeyAuth(token=hash_token(user_key))) setattr(litellm.proxy.proxy_server, "user_api_key_cache", user_api_key_cache) setattr(litellm.proxy.proxy_server, "master_key", "sk-1234") @@ -1235,9 +1196,7 @@ async def test_user_api_key_from_query_param(): "query_string": f"alt=sse&key={user_key}".encode(), } ) - request._url = URL( - url=f"/v1beta/models/gemini:streamGenerateContent?alt=sse&key={user_key}" - ) + request._url = URL(url=f"/v1beta/models/gemini:streamGenerateContent?alt=sse&key={user_key}") async def return_body(): return b"{}" @@ -1246,3 +1205,592 @@ async def test_user_api_key_from_query_param(): valid_token = await user_api_key_auth(request=request, api_key="") assert valid_token.token == hash_token(user_key) + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "user_id,user_model_max_budget,expected_calls", + [ + ("u-1", {"gpt-4": {"budget_limit": 1.0, "time_period": "1mo"}}, 1), + ("u-1", {}, 0), + ("u-1", None, 0), + (None, {"gpt-4": {"budget_limit": 1.0, "time_period": "1mo"}}, 0), + ], + ids=["enforced", "empty_budget", "no_budget", "no_user_id"], +) +async def test_check_user_model_budget(user_id, user_model_max_budget, expected_calls): + """ + An internal user's model_max_budget must reach the limiter. Before this it was + stored on LiteLLM_UserTable, accepted by /user/new and /user/update, and read + by nothing, so a user-level per-model budget never blocked anything. + """ + from litellm.proxy.auth.user_api_key_auth import _check_user_model_budget + + calls = [] + + class _Limiter: + async def is_user_within_model_budget(self, user_id, user_model_max_budget, model): + calls.append((user_id, user_model_max_budget, model)) + return True + + valid_token = UserAPIKeyAuth( + token="hash", + user_id=user_id, + user_model_max_budget=user_model_max_budget, + ) + await _check_user_model_budget( + valid_token=valid_token, + model_max_budget_limiter=_Limiter(), + models=["gpt-4"], + ) + assert len(calls) == expected_calls + if expected_calls: + assert calls[0] == ("u-1", user_model_max_budget, "gpt-4") + + +@pytest.mark.asyncio +async def test_user_model_max_budget_is_threaded_onto_the_auth_object(): + """ + The limiter can only enforce what auth carries. Regression for the user row's + model_max_budget being dropped on the way into UserAPIKeyAuth. + """ + from datetime import datetime + + from litellm.proxy.auth.user_api_key_auth import _return_user_api_key_auth_obj + + budget = {"gpt-4": {"budget_limit": 1.0, "time_period": "1mo"}} + user_obj = LiteLLM_UserTable( + user_id="u-1", + max_budget=None, + spend=0.0, + user_email=None, + models=[], + model_max_budget=budget, + ) + + auth_obj = await _return_user_api_key_auth_obj( + user_obj=user_obj, + api_key="sk-1234", + parent_otel_span=None, + valid_token_dict={"token": "hash"}, + route="/chat/completions", + start_time=datetime.now(), + user_role=LitellmUserRoles.INTERNAL_USER, + ) + assert auth_obj.user_model_max_budget == budget + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "over_budget,expect_refusal", + [(True, True), (False, False)], + ids=["over_budget_is_refused", "under_budget_is_served"], +) +async def test_user_model_budget_is_enforced_through_user_api_key_auth(over_budget, expect_refusal): + """ + Drive the real auth entry point, not the helper. + + The user's model_max_budget lives on the user row, and the joint + verification-token view auth builds its token from does not carry it. A test + that only exercises the helper passes while the whole path is inert, so this + one goes through user_api_key_auth with a key that has no per-model budget of + its own and asserts the USER's budget decides the outcome. + """ + from fastapi import Request + from starlette.datastructures import URL + + from litellm.proxy._types import LiteLLM_UserTable, Litellm_EntityType, UserAPIKeyAuth + from litellm.proxy.auth.user_api_key_auth import user_api_key_auth + from litellm.proxy.hooks.model_max_budget_limiter import model_budget_spend_cache_key + from litellm.proxy.proxy_server import ( + hash_token, + model_max_budget_limiter, + user_api_key_cache, + ) + + user_id = "user-model-budget" + model = "gpt-4o" + key = "sk-user-model-budget" + hashed = hash_token(key) + user_model_max_budget = {model: {"budget_limit": 1.0, "time_period": "1mo"}} + + setattr(litellm.proxy.proxy_server, "user_api_key_cache", user_api_key_cache) + setattr(litellm.proxy.proxy_server, "master_key", "sk-1234") + setattr(litellm.proxy.proxy_server, "prisma_client", "present") + + await user_api_key_cache.async_set_cache( + key=hashed, + value=UserAPIKeyAuth(token=hashed, user_id=user_id, models=[], model_max_budget={}), + model_type=UserAPIKeyAuth, + ) + await model_max_budget_limiter.dual_cache.async_set_cache( + key=model_budget_spend_cache_key( + entity_type=Litellm_EntityType.USER, + entity_id=user_id, + budget_model=model, + budget_duration="1mo", + ), + value=5.0 if over_budget else 0.25, + ttl=600, + ) + + request = Request(scope={"type": "http"}) + request._url = URL(url="/chat/completions") + + async def return_body(): + return f'{{"model": "{model}"}}'.encode() + + request.body = return_body + + async def fake_get_user_object(**kwargs): + return LiteLLM_UserTable( + user_id=user_id, + max_budget=None, + spend=0.0, + user_email=None, + models=[], + model_max_budget=user_model_max_budget, + ) + + with patch( + "litellm.proxy.auth.user_api_key_auth.get_user_object", + new=fake_get_user_object, + ): + if expect_refusal: + with pytest.raises(Exception) as exc: + await user_api_key_auth(request=request, api_key="Bearer " + key) + assert "budget" in str(exc.value).lower() + assert user_id in str(exc.value) + else: + result = await user_api_key_auth(request=request, api_key="Bearer " + key) + # The budget must also reach the token, or the post-call increment + # has nothing to charge and the counter never grows. + assert result.user_model_max_budget == user_model_max_budget + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "over_budget,expect_refusal", + [(True, True), (False, False)], + ids=["over_budget_is_refused", "under_budget_is_served"], +) +async def test_jwt_user_model_budget_is_enforced_before_the_jwt_path_returns(over_budget, expect_refusal): + """ + JWT auth returns its own token instead of falling through to the + virtual-key budget checks, so the user's per-model budget has to be enforced + on that path explicitly. + + The dangerous shape is not "no tracking": the post-call increment charges the + JWT user's counter either way, so without this check the counter grows and + nothing ever reads it, which looks enforced and is not. + """ + from litellm.proxy._types import Litellm_EntityType, UserAPIKeyAuth + from litellm.proxy.auth.user_api_key_auth import _check_user_model_budget + from litellm.proxy.hooks.model_max_budget_limiter import model_budget_spend_cache_key + from litellm.proxy.proxy_server import model_max_budget_limiter + + user_id = "jwt-user-model-budget" + model = "gpt-4o" + user_model_max_budget = {model: {"budget_limit": 1.0, "time_period": "1mo"}} + + await model_max_budget_limiter.dual_cache.async_set_cache( + key=model_budget_spend_cache_key( + entity_type=Litellm_EntityType.USER, + entity_id=user_id, + budget_model=model, + budget_duration="1mo", + ), + value=5.0 if over_budget else 0.25, + ttl=600, + ) + + # The token the JWT branch builds and returns. + valid_token = UserAPIKeyAuth( + api_key=None, + user_id=user_id, + user_model_max_budget=user_model_max_budget, + ) + + if expect_refusal: + with pytest.raises(litellm.BudgetExceededError) as exc: + await _check_user_model_budget( + valid_token=valid_token, + model_max_budget_limiter=model_max_budget_limiter, + models=[model], + ) + assert exc.value.entity_type == Litellm_EntityType.USER.value + else: + await _check_user_model_budget( + valid_token=valid_token, + model_max_budget_limiter=model_max_budget_limiter, + models=[model], + ) + + +def test_jwt_path_enforces_the_user_model_budget_before_returning(): + """ + The JWT branch returns early, so the enforcement call has to sit before that + return rather than in the virtual-key block. Assert on the call graph, since + a helper-level test passes whether or not the JWT path ever calls it. + """ + import ast + import inspect + import textwrap + + from litellm.proxy.auth import user_api_key_auth as auth_module + + tree = ast.parse(textwrap.dedent(inspect.getsource(auth_module._user_api_key_auth_builder))) + + def calls_before_each_return(node): + seen_check = [] + for child in ast.walk(node): + if isinstance(child, ast.Call): + fn = child.func + name = getattr(fn, "id", None) or getattr(fn, "attr", None) + if name == "_check_user_model_budget": + seen_check.append(child.lineno) + return seen_check + + check_lines = calls_before_each_return(tree) + assert check_lines, "_user_api_key_auth_builder never enforces the user model budget" + + jwt_returns = [ + n.lineno + for n in ast.walk(tree) + if isinstance(n, ast.Return) and isinstance(n.value, ast.Call) and getattr(n.value.func, "id", None) == "cast" + ] + assert jwt_returns, "expected the JWT branch's `return cast(UserAPIKeyAuth, valid_token)`" + assert any(check < jwt_return for check in check_lines for jwt_return in jwt_returns), ( + "the user model-budget check must run before the JWT branch returns" + ) + + +def test_every_jwt_branch_carries_the_user_model_budget(): + """ + Each JWT branch that builds or replaces `valid_token` has to put the user's + model budget on it, or the enforcement call a few lines later has nothing to + read and silently admits the request. + + The auto-register branch is the one that regressed: it REPLACES the token + built above it with a key-scoped one whose columns carry no user budget. + """ + import ast + import inspect + import textwrap + + from litellm.proxy.auth import user_api_key_auth as auth_module + + tree = ast.parse(textwrap.dedent(inspect.getsource(auth_module._user_api_key_auth_builder))) + + assignments = [ + node + for node in ast.walk(tree) + if isinstance(node, ast.Assign) + and any(isinstance(t, ast.Attribute) and t.attr == "user_model_max_budget" for t in node.targets) + ] + targets = { + t.value.id + for node in assignments + for t in node.targets + if isinstance(t, ast.Attribute) and isinstance(t.value, ast.Name) + } + assert "auto_registered" in targets, ( + f"the auto-registered JWT token must carry the user's model budget; only these are populated: {sorted(targets)}" + ) + assert "valid_token" in targets, "the virtual-key path must carry the user's model budget" + + +@pytest.mark.asyncio +async def test_user_budget_lookup_tolerates_an_unreadable_user(): + """ + `get_user_object(user_id_upsert=False)` raises a bare Exception when the row + is simply ABSENT, which is the ordinary state for a custom-auth deployment + that never writes users to the proxy DB. Refusing on that exception would + turn "no user row" into a 4xx for every such request, and a transient DB + blip into a full outage. + + The virtual-key path makes the same call and swallows the same exception + ("Unable to get user from db/cache. Setting user_obj to None"), so this is + the established contract, not a shortcut. There is also nothing to enforce: + the budget being looked up lives on the row that could not be read. + """ + from litellm.caching.dual_cache import DualCache + from litellm.proxy.auth.user_api_key_auth import _read_user_model_max_budget + + prisma_client = MagicMock() + + with patch( + "litellm.proxy.auth.user_api_key_auth.get_user_object", + new=AsyncMock(side_effect=Exception("No user table row")), + ): + budget = await _read_user_model_max_budget( + user_id="user-with-no-row", + prisma_client=prisma_client, + user_api_key_cache=DualCache(), + parent_otel_span=None, + proxy_logging_obj=MagicMock(), + ) + + assert budget is None + + +@pytest.mark.asyncio +async def test_user_budget_lookup_is_also_unenforced_when_the_database_is_down(): + """ + KNOWN LIMITATION, pinned deliberately rather than discovered later. + + `get_user_object` cannot tell "row absent" from "database unreachable": the + absent case raises inside its own try (auth_checks.py:2177) and the handler + at :2213 rewrites every exception into the same + `ValueError("User doesn't exist in db...")`. A connection error, a query + timeout and a malformed row all reach us as that one type and message. + + So tolerating the absent case, which the test above requires, unavoidably + tolerates an outage too, and a user who DOES have a per-model budget goes + unenforced while the DB is unreachable. This is pre-existing behaviour of + `get_user_object` that the virtual-key path inherits identically; it is not + introduced here. Distinguishing them needs a dedicated exception type for + the absent case and a change to both auth paths. + """ + from litellm.caching.dual_cache import DualCache + from litellm.proxy.auth.user_api_key_auth import _read_user_model_max_budget + + db_down = ValueError("User doesn't exist in db. 'user_id'=u-1. Got error - Connection refused") + + with patch( + "litellm.proxy.auth.user_api_key_auth.get_user_object", + new=AsyncMock(side_effect=db_down), + ): + budget = await _read_user_model_max_budget( + user_id="u-1", + prisma_client=MagicMock(), + user_api_key_cache=DualCache(), + parent_otel_span=None, + proxy_logging_obj=MagicMock(), + ) + + assert budget is None + + +@pytest.mark.asyncio +async def test_user_budget_lookup_returns_the_budget_when_the_row_reads(): + """Positive control: the tolerance above must not be swallowing every result.""" + from litellm.caching.dual_cache import DualCache + from litellm.proxy.auth.user_api_key_auth import _read_user_model_max_budget + + stored = {"claude-opus-4-8": {"budget_limit": 1.0, "time_period": "18h"}} + user_obj = MagicMock() + user_obj.model_max_budget = stored + + with patch( + "litellm.proxy.auth.user_api_key_auth.get_user_object", + new=AsyncMock(return_value=user_obj), + ): + budget = await _read_user_model_max_budget( + user_id="user-1", + prisma_client=MagicMock(), + user_api_key_cache=DualCache(), + parent_otel_span=None, + proxy_logging_obj=MagicMock(), + ) + + assert budget == stored + + +def test_zero_cost_models_skip_the_user_budget_check_on_every_path(): + """ + `skip_budget_checks` is computed per request for zero-cost models, and the + JWT branch logs "Skipping all budget checks" when it is set. Any enforcement + call that ignores it makes the same request behave differently depending on + whether the caller used a JWT or a virtual key, and makes that log a lie. + + Structural rather than behavioural on purpose: the defect is a call site + sitting outside a guard, and driving both auth paths to a zero-cost model + would prove it for the two requests exercised rather than for every site. + """ + import ast + import inspect + import textwrap + + from litellm.proxy.auth import user_api_key_auth as auth_module + + tree = ast.parse(textwrap.dedent(inspect.getsource(auth_module._user_api_key_auth_builder))) + + def guarded_by_skip(node: ast.AST, target: ast.AST) -> bool: + for parent in ast.walk(node): + if not isinstance(parent, ast.If): + continue + test = parent.test + is_skip_guard = ( + isinstance(test, ast.UnaryOp) + and isinstance(test.op, ast.Not) + and isinstance(test.operand, ast.Name) + and test.operand.id == "skip_budget_checks" + ) + if is_skip_guard and any(sub is target for sub in ast.walk(parent)): + return True + return False + + calls = [ + node + for node in ast.walk(tree) + if isinstance(node, ast.Call) and isinstance(node.func, ast.Name) and node.func.id == "_check_user_model_budget" + ] + assert len(calls) == 2, f"expected the JWT and virtual-key call sites, found {len(calls)}" + + unguarded = [c for c in calls if not guarded_by_skip(tree, c)] + assert not unguarded, ( + f"{len(unguarded)} _check_user_model_budget call(s) run even when " + "skip_budget_checks is set, so a zero-cost model is enforced on one auth path and not the other" + ) + + +def test_custom_auth_also_skips_budget_checks_for_zero_cost_models(): + """ + The custom-auth helper runs its own key, user and end-user per-model budget + checks. If it does not honour the zero-cost skip that the JWT and + virtual-key paths honour, the same free request is refused under one auth + method and served under the others. + + Asserted structurally, on the same reasoning as the sibling test: the defect + is a check sitting outside a guard, and it must hold for checks added later + rather than only for whichever request a behavioural test happened to drive. + """ + import ast + import inspect + import textwrap + + from litellm.proxy.auth import user_api_key_auth as auth_module + + src = textwrap.dedent(inspect.getsource(auth_module._run_post_custom_auth_checks)) + tree = ast.parse(src) + + assert "skip_budget_checks" in src, "the custom-auth path never computes the zero-cost skip flag" + + budget_calls = ( + "_check_key_model_budget_with_fallback", + "_check_user_model_budget", + "is_end_user_within_model_budget", + ) + + def guarding_ifs(target: ast.AST) -> list[ast.If]: + return [ + node for node in ast.walk(tree) if isinstance(node, ast.If) and any(sub is target for sub in ast.walk(node)) + ] + + def mentions_skip(node: ast.If) -> bool: + return any(isinstance(sub, ast.Name) and sub.id == "skip_budget_checks" for sub in ast.walk(node.test)) + + for call_name in budget_calls: + calls = [ + node + for node in ast.walk(tree) + if isinstance(node, ast.Call) + and ( + (isinstance(node.func, ast.Name) and node.func.id == call_name) + or (isinstance(node.func, ast.Attribute) and node.func.attr == call_name) + ) + ] + assert calls, f"{call_name} is no longer called here; update this invariant" + for call in calls: + assert any(mentions_skip(node) for node in guarding_ifs(call)), ( + f"{call_name} runs even for a zero-cost model, so custom auth refuses " + "requests the JWT and virtual-key paths serve" + ) + + +def test_custom_auth_attaches_the_user_budget_even_when_it_does_not_enforce(): + """ + The post-call spend hook reads `user_model_max_budget` off the token, so the + attach has to happen whether or not THIS request was enforceable. Gating it + on the same condition as the check leaves the user's counter uncharged for + every request with no resolvable model or a zero-cost one, which is exactly + the untracked-spend defect this PR fixes. + + Structural, because the failure is an assignment sitting inside a guard. + """ + import ast + import inspect + import textwrap + + from litellm.proxy.auth import user_api_key_auth as auth_module + + tree = ast.parse(textwrap.dedent(inspect.getsource(auth_module._run_post_custom_auth_checks))) + + attaches = [ + node + for node in ast.walk(tree) + if isinstance(node, ast.Assign) + and any(isinstance(t, ast.Attribute) and t.attr == "user_model_max_budget" for t in node.targets) + ] + assert attaches, "custom auth no longer attaches the user budget at all" + + for attach in attaches: + enclosing_ifs = [ + node for node in ast.walk(tree) if isinstance(node, ast.If) and any(sub is attach for sub in ast.walk(node)) + ] + assert not enclosing_ifs, ( + "the user budget is attached inside a conditional, so the spend hook " + "cannot charge the user counter whenever that condition is false" + ) + + +def test_mapped_key_jwt_falls_through_to_the_shared_user_budget_attach(): + """ + A JWT that maps to an existing virtual key resolves through the resolver + store, which builds the token from the KEY row alone and therefore carries + no user-level per-model budget. That branch sets `do_standard_jwt_auth = + False` precisely so it falls through to the shared virtual-key checks, where + the user row is loaded and its budget copied onto the token. + + Reviewed as a bypass three times, so the two halves it depends on are pinned + here: the branch must not return before the shared block, and the shared + block must copy the user row's budget onto the token. Structural on purpose, + because the claim is about control flow reaching a statement, and it has to + hold for branches added later rather than for one mocked request. + """ + import ast + import inspect + import textwrap + + from litellm.proxy.auth import user_api_key_auth as auth_module + + tree = ast.parse(textwrap.dedent(inspect.getsource(auth_module._user_api_key_auth_builder))) + + # Half one: the shared block copies the user row's budget onto the token. + copies_user_row = [ + node + for node in ast.walk(tree) + if isinstance(node, ast.Assign) + and any(isinstance(t, ast.Attribute) and t.attr == "user_model_max_budget" for t in node.targets) + and any(isinstance(v, ast.Attribute) and v.attr == "model_max_budget" for v in ast.walk(node.value)) + ] + assert copies_user_row, ( + "nothing copies the user row's model_max_budget onto the token, so a mapped-key " + "JWT reaches enforcement carrying the key's columns only" + ) + + # Half two: the mapped-key branch does not return before reaching it. + disables_standard_auth = [ + node + for node in ast.walk(tree) + if isinstance(node, ast.Assign) + and any(isinstance(t, ast.Name) and t.id == "do_standard_jwt_auth" for t in node.targets) + and isinstance(node.value, ast.Constant) + and node.value.value is False + ] + assert len(disables_standard_auth) == 1, "expected exactly one mapped-key branch" + marker = disables_standard_auth[0] + + enclosing = [ + node for node in ast.walk(tree) if isinstance(node, ast.If) and any(sub is marker for sub in node.body) + ] + assert enclosing, "could not locate the mapped-key branch body" + + returns_after = [ + node for node in ast.walk(enclosing[0]) if isinstance(node, ast.Return) and node.lineno > marker.lineno + ] + assert not returns_after, ( + "the mapped-key branch returns before the shared virtual-key checks, so the " + "user's per-model budget is never attached and never enforced" + ) diff --git a/tests/test_litellm/llms/anthropic/experimental_pass_through/context_management/test_compact.py b/tests/test_litellm/llms/anthropic/experimental_pass_through/context_management/test_compact.py index 6cc1d9e5add..28c82fdf528 100644 --- a/tests/test_litellm/llms/anthropic/experimental_pass_through/context_management/test_compact.py +++ b/tests/test_litellm/llms/anthropic/experimental_pass_through/context_management/test_compact.py @@ -1202,6 +1202,8 @@ def _fake_user_api_key_auth( model_max_budget=None, end_user_model_max_budget=None, end_user_id=None, + user_model_max_budget=None, + user_id=None, token=None, ): """Build a minimal stand-in for ``UserAPIKeyAuth`` with just the fields @@ -1220,6 +1222,8 @@ def _fake_user_api_key_auth( auth.model_max_budget = model_max_budget auth.end_user_model_max_budget = end_user_model_max_budget auth.end_user_id = end_user_id + auth.user_model_max_budget = user_model_max_budget + auth.user_id = user_id auth.token = token return auth @@ -1548,6 +1552,78 @@ async def test_summary_model_denied_when_key_over_model_budget(): assert result.applied_edits[0].get("error") == "summary_model_budget_exceeded" +async def test_summary_model_denied_when_user_over_model_budget(): + """Internal-user per-model budget is enforced for the summary subrequest too. + + This file propagates `user_api_key_user_model_max_budget` into the summary + subrequest's metadata, so its spend charges the user's counter. Enforcing + only the key and end-user scopes would let compaction increment a counter it + can never be refused by, which is the asymmetry this PR exists to remove. + """ + import litellm + + messages = _simple_messages() + mock_call = AsyncMock(return_value=_make_mock_response("x")) + + auth = _fake_user_api_key_auth( + key_models=["all-proxy-models"], + user_model_max_budget={"claude-haiku-4-5": {"budget_limit": 5}}, + user_id="user-over-budget", + token="hashed-token", + ) + + limiter = MagicMock() + limiter.is_user_within_model_budget = AsyncMock( + side_effect=litellm.BudgetExceededError( + message="over budget", current_cost=10, max_budget=5 + ) + ) + + with ( + patch( + "litellm.llms.anthropic.experimental_pass_through.context_management.editors.compact._read_summary_model_setting", + return_value="claude-haiku-4-5", + ), + patch("litellm.token_counter", return_value=200_000), + patch( + "litellm.llms.anthropic.experimental_pass_through.context_management.editors.compact._call_summary_model", + mock_call, + ), + patch("litellm.proxy.proxy_server.model_max_budget_limiter", limiter), + ): + result = await apply_compact_20260112( + model=MODEL, + messages=messages, + tools=None, + system=None, + edit_spec=_EDIT_SPEC_DEFAULT, + user_api_key_auth=auth, + ) + + mock_call.assert_not_awaited() + assert result.applied_edits[0].get("error") == "summary_model_budget_exceeded" + + # The limiter is a mock, so it would accept any kwargs. Pin the call shape and + # check it against the real method, or a rename there would keep this test + # green while breaking compaction in production. + limiter.is_user_within_model_budget.assert_awaited_once_with( + user_id="user-over-budget", + user_model_max_budget={"claude-haiku-4-5": {"budget_limit": 5}}, + model="claude-haiku-4-5", + ) + import inspect + + from litellm.proxy.hooks.model_max_budget_limiter import ( + _PROXY_VirtualKeyModelMaxBudgetLimiter, + ) + + real_params = inspect.signature( + _PROXY_VirtualKeyModelMaxBudgetLimiter.is_user_within_model_budget + ).parameters + for kwarg in ("user_id", "user_model_max_budget", "model"): + assert kwarg in real_params, f"compact.py passes {kwarg}=, which the limiter no longer accepts" + + async def test_summary_model_denied_when_end_user_over_model_budget(): """End-user per-model budget is enforced for the summary subrequest too.""" import litellm diff --git a/tests/test_litellm/proxy/management_endpoints/test_internal_user_endpoints.py b/tests/test_litellm/proxy/management_endpoints/test_internal_user_endpoints.py index 8a036d7e62f..da51d513b39 100644 --- a/tests/test_litellm/proxy/management_endpoints/test_internal_user_endpoints.py +++ b/tests/test_litellm/proxy/management_endpoints/test_internal_user_endpoints.py @@ -2975,6 +2975,7 @@ async def test_user_info_v2_response_shape(mocker): "updated_at": datetime(2024, 6, 1, tzinfo=timezone.utc), "sso_user_id": None, "teams": ["team-a", "team-b"], + "model_max_budget": {"gpt-3.5-turbo": {"budget_limit": 5.0, "time_period": "30d"}}, } async def mock_find_unique(*args, **kwargs): @@ -3018,9 +3019,20 @@ async def test_user_info_v2_response_shape(mocker): "sso_user_id", "teams", "object_permission", + "model_max_budget", + "model_max_budget_usage", } assert set(response_dict.keys()) == expected_fields + # The dashboard's user edit form hydrates its per-model budget rows from + # these two, so dropping them makes a save replace the user's budgets. + assert response_dict["model_max_budget"] == { + "gpt-3.5-turbo": {"budget_limit": 5.0, "time_period": "30d"} + } + assert response_dict["model_max_budget_usage"] == { + "gpt-3.5-turbo": {"current_spend": 0.0, "budget_limit": 5.0, "time_period": "30d"} + } + # Verify teams is a list of strings (team IDs), not team objects assert isinstance(response.teams, list) assert all(isinstance(t, str) for t in response.teams) @@ -4150,3 +4162,66 @@ async def test_user_info_v2_returns_the_mcp_entitlement(mocker): assert response.object_permission.mcp_tool_permissions == { "github": ["list_issues"] } + + +@pytest.mark.asyncio +@pytest.mark.parametrize( + "model_max_budget,expected_written", + [ + ( + {"claude-opus-4-8": {"budget_limit": 200.0, "time_period": "1mo"}}, + '{"claude-opus-4-8": {"budget_limit": 200.0, "time_period": "1mo"}}', + ), + (None, None), + ({}, None), + ], + ids=["supplied", "omitted", "empty"], +) +async def test_user_new_persists_model_max_budget( + monkeypatch, model_max_budget, expected_written +): + """ + /user/new used to echo model_max_budget back while writing {} to the user row, + so a per-model budget looked configured and was read by nothing. + + The omitted/empty cases are the other half: SSO and default-key callers reach + generate_key_helper_fn with no budget, and writing "{}" for them would clear + an existing user's budgets. + """ + from litellm.proxy.management_endpoints import key_management_endpoints + + captured = {} + + class _FakeUserRow: + models = [] + + class _FakePrisma: + async def insert_data(self, data, table_name): + if table_name == "user": + captured["user_data"] = dict(data) + return _FakeUserRow() + captured["key_data"] = dict(data) + return SimpleNamespace( + token=data.get("token"), + litellm_budget_table=None, + created_at=None, + updated_at=None, + ) + + async def get_data(self, *args, **kwargs): + return None + + import litellm.proxy.proxy_server as proxy_server + + monkeypatch.setattr(proxy_server, "prisma_client", _FakePrisma(), raising=False) + # model_max_budget is an enterprise feature; without this the call is rejected + # before it ever reaches the write this test is about. + monkeypatch.setattr(proxy_server, "premium_user", True, raising=False) + + await key_management_endpoints.generate_key_helper_fn( + request_type="user", + user_id="u-1", + model_max_budget=model_max_budget, + ) + + assert captured["user_data"].get("model_max_budget") == expected_written diff --git a/tests/test_litellm/proxy/management_endpoints/test_key_management_endpoints.py b/tests/test_litellm/proxy/management_endpoints/test_key_management_endpoints.py index 069cfa01178..fff6368cfc6 100644 --- a/tests/test_litellm/proxy/management_endpoints/test_key_management_endpoints.py +++ b/tests/test_litellm/proxy/management_endpoints/test_key_management_endpoints.py @@ -13507,10 +13507,15 @@ async def test_info_key_fn_includes_model_max_budget_usage(monkeypatch): mock_prisma_client = AsyncMock() monkeypatch.setattr("litellm.proxy.proxy_server.prisma_client", mock_prisma_client) mock_user_api_key_cache = AsyncMock() - mock_user_api_key_cache.async_get_cache = AsyncMock(return_value=0.23) monkeypatch.setattr( "litellm.proxy.proxy_server.user_api_key_cache", mock_user_api_key_cache ) + # A real cache seeded at the real counter key: the spend only comes back if + # the endpoint computed virtual_key_spend:hashed_token_budget_test:gpt-4o:1d. + monkeypatch.setattr( + "litellm.proxy.proxy_server.model_max_budget_limiter.dual_cache", + await _budget_cache({"virtual_key_spend:hashed_token_budget_test:gpt-4o:1d": 0.23}), + ) mock_key_info = MagicMock(spec=LiteLLM_VerificationToken) mock_key_info.token = test_key_token @@ -13568,6 +13573,10 @@ async def test_info_key_fn_no_model_max_budget_skips_usage(monkeypatch): monkeypatch.setattr( "litellm.proxy.proxy_server.user_api_key_cache", mock_user_api_key_cache ) + monkeypatch.setattr( + "litellm.proxy.proxy_server.model_max_budget_limiter.dual_cache", + mock_user_api_key_cache, + ) mock_key_info = MagicMock(spec=LiteLLM_VerificationToken) mock_key_info.token = test_key_token @@ -13621,10 +13630,15 @@ async def test_info_key_fn_v2_includes_model_max_budget_usage(monkeypatch): mock_prisma_client = AsyncMock() monkeypatch.setattr("litellm.proxy.proxy_server.prisma_client", mock_prisma_client) mock_user_api_key_cache = AsyncMock() - mock_user_api_key_cache.async_get_cache = AsyncMock(return_value=0.55) monkeypatch.setattr( "litellm.proxy.proxy_server.user_api_key_cache", mock_user_api_key_cache ) + # A real cache seeded at the real counter key: the spend only comes back if + # the endpoint computed virtual_key_spend:hashed_token_v2_test:gpt-4o:7d. + monkeypatch.setattr( + "litellm.proxy.proxy_server.model_max_budget_limiter.dual_cache", + await _budget_cache({"virtual_key_spend:hashed_token_v2_test:gpt-4o:7d": 0.55}), + ) mock_key = MagicMock(spec=LiteLLM_VerificationToken) mock_key.token = test_key_token @@ -13680,10 +13694,15 @@ async def test_info_key_fn_budget_table_fallback(monkeypatch): mock_prisma_client = AsyncMock() monkeypatch.setattr("litellm.proxy.proxy_server.prisma_client", mock_prisma_client) mock_user_api_key_cache = AsyncMock() - mock_user_api_key_cache.async_get_cache = AsyncMock(return_value=1.20) monkeypatch.setattr( "litellm.proxy.proxy_server.user_api_key_cache", mock_user_api_key_cache ) + # A real cache seeded at the real counter key: the spend only comes back if + # the endpoint computed virtual_key_spend:hashed_token_budget_table_test:bedrock/anthropic.claude-opus-4:30d. + monkeypatch.setattr( + "litellm.proxy.proxy_server.model_max_budget_limiter.dual_cache", + await _budget_cache({"virtual_key_spend:hashed_token_budget_table_test:bedrock/anthropic.claude-opus-4:30d": 1.20}), + ) mock_key_info = MagicMock(spec=LiteLLM_VerificationToken) mock_key_info.token = test_key_token @@ -13748,10 +13767,15 @@ async def test_info_key_fn_v2_budget_table_fallback(monkeypatch): mock_prisma_client = AsyncMock() monkeypatch.setattr("litellm.proxy.proxy_server.prisma_client", mock_prisma_client) mock_user_api_key_cache = AsyncMock() - mock_user_api_key_cache.async_get_cache = AsyncMock(return_value=2.50) monkeypatch.setattr( "litellm.proxy.proxy_server.user_api_key_cache", mock_user_api_key_cache ) + # A real cache seeded at the real counter key: the spend only comes back if + # the endpoint computed virtual_key_spend:hashed_token_v2_bt_test:bedrock/anthropic.claude-opus-4:30d. + monkeypatch.setattr( + "litellm.proxy.proxy_server.model_max_budget_limiter.dual_cache", + await _budget_cache({"virtual_key_spend:hashed_token_v2_bt_test:bedrock/anthropic.claude-opus-4:30d": 2.50}), + ) mock_key = MagicMock(spec=LiteLLM_VerificationToken) mock_key.token = test_key_token @@ -13793,8 +13817,13 @@ async def test_info_key_fn_v2_budget_table_fallback(monkeypatch): @pytest.mark.asyncio -async def test_info_key_fn_provider_prefix_spend_fallback(monkeypatch): - """Cached spend for 'gpt-4o' matches budget key 'openai/gpt-4o' via suffix match.""" +async def test_info_key_fn_reads_the_configured_budget_model_key(monkeypatch): + """/key/info reads the one counter enforcement reads: the configured budget model. + + It used to probe a second, provider-stripped key because the counter was + written under the request model instead, which is what let a key report zero + usage while being blocked at 429. + """ from unittest.mock import AsyncMock, MagicMock from litellm.proxy._types import LiteLLM_VerificationToken @@ -13808,10 +13837,15 @@ async def test_info_key_fn_provider_prefix_spend_fallback(monkeypatch): mock_prisma_client = AsyncMock() monkeypatch.setattr("litellm.proxy.proxy_server.prisma_client", mock_prisma_client) mock_user_api_key_cache = AsyncMock() - mock_user_api_key_cache.async_get_cache = AsyncMock(side_effect=[None, 0.75]) monkeypatch.setattr( "litellm.proxy.proxy_server.user_api_key_cache", mock_user_api_key_cache ) + # A real cache seeded at the real counter key: the spend only comes back if + # the endpoint computed virtual_key_spend:hashed_token_prefix_test:openai/gpt-4o:7d. + monkeypatch.setattr( + "litellm.proxy.proxy_server.model_max_budget_limiter.dual_cache", + await _budget_cache({"virtual_key_spend:hashed_token_prefix_test:openai/gpt-4o:7d": 0.75}), + ) mock_key_info = MagicMock(spec=LiteLLM_VerificationToken) mock_key_info.token = test_key_token @@ -13847,7 +13881,22 @@ async def test_info_key_fn_provider_prefix_spend_fallback(monkeypatch): assert "model_max_budget_usage" in result["info"] usage = result["info"]["model_max_budget_usage"] assert usage["openai/gpt-4o"]["current_spend"] == 0.75 - assert mock_user_api_key_cache.async_get_cache.await_count == 2 + + +async def _budget_cache(seeded): + """A real DualCache holding spend at the given LITERAL counter keys. + + The keys are spelled out in full on purpose. Seeding via + model_budget_spend_cache_key would move the seed and the read together, so + any change to the key format would still match itself and these tests could + never fail, which is the exact bug they exist to catch. + """ + from litellm.caching.caching import DualCache + + cache = DualCache() + for key, spend in seeded.items(): + await cache.async_set_cache(key, spend) + return cache @pytest.mark.asyncio @@ -13872,19 +13921,16 @@ async def test_build_model_max_budget_usage_reads_current_cache_window(): _build_model_max_budget_usage, ) - mock_user_api_key_cache = AsyncMock() - mock_user_api_key_cache.async_get_cache = AsyncMock(return_value=0.30) + cache = await _budget_cache({"virtual_key_spend:some-hash:gpt-4o:30d": 0.30}) result = await _build_model_max_budget_usage( api_key_hash="some-hash", model_max_budget={"gpt-4o": {"budget_limit": 1.0, "time_period": "30d"}}, - user_api_key_cache=mock_user_api_key_cache, + user_api_key_cache=cache, ) + # 0.30 comes back only if the key matched virtual_key_spend:some-hash:gpt-4o:30d. assert result["gpt-4o"]["current_spend"] == 0.30 - mock_user_api_key_cache.async_get_cache.assert_awaited_once_with( - key="virtual_key_spend:some-hash:gpt-4o:30d" - ) @pytest.mark.asyncio @@ -13915,8 +13961,7 @@ async def test_build_model_max_budget_usage_skips_model_without_duration(): _build_model_max_budget_usage, ) - mock_user_api_key_cache = AsyncMock() - mock_user_api_key_cache.async_get_cache = AsyncMock(return_value=0.10) + cache = await _budget_cache({"virtual_key_spend:some-hash:gpt-4o:1d": 0.10}) result = await _build_model_max_budget_usage( api_key_hash="some-hash", @@ -13924,11 +13969,10 @@ async def test_build_model_max_budget_usage_skips_model_without_duration(): "gpt-4o": {"budget_limit": 1.0, "time_period": "1d"}, "gpt-3.5-turbo": {"budget_limit": 0.5}, }, - user_api_key_cache=mock_user_api_key_cache, + user_api_key_cache=cache, ) - assert "gpt-4o" in result + assert result["gpt-4o"]["current_spend"] == 0.10 assert "gpt-3.5-turbo" not in result - assert mock_user_api_key_cache.async_get_cache.await_count == 1 @pytest.mark.asyncio @@ -13961,8 +14005,7 @@ async def test_build_model_max_budget_usage_invalid_budget_config_skipped(): _build_model_max_budget_usage, ) - mock_user_api_key_cache = AsyncMock() - mock_user_api_key_cache.async_get_cache = AsyncMock(return_value=0.20) + cache = await _budget_cache({"virtual_key_spend:some-hash:gpt-3.5-turbo:7d": 0.20}) result = await _build_model_max_budget_usage( api_key_hash="some-hash", @@ -13970,32 +14013,35 @@ async def test_build_model_max_budget_usage_invalid_budget_config_skipped(): "gpt-4o": {"max_budget": "not-a-number", "budget_duration": "1d"}, "gpt-3.5-turbo": {"budget_limit": 0.5, "time_period": "7d"}, }, - user_api_key_cache=mock_user_api_key_cache, + user_api_key_cache=cache, ) assert "gpt-4o" not in result - assert "gpt-3.5-turbo" in result - assert mock_user_api_key_cache.async_get_cache.await_count == 1 + assert result["gpt-3.5-turbo"]["current_spend"] == 0.20 @pytest.mark.asyncio -async def test_build_model_max_budget_usage_provider_prefix_cache_fallback(): +async def test_build_model_max_budget_usage_reads_only_the_configured_model_key(): + """One lookup, at the configured budget model. + + The counter is written under the name the operator configured, so probing a + provider-stripped variant would read a key nothing writes. + """ from unittest.mock import AsyncMock from litellm.proxy.management_endpoints.key_management_endpoints import ( _build_model_max_budget_usage, ) - mock_user_api_key_cache = AsyncMock() - mock_user_api_key_cache.async_get_cache = AsyncMock(side_effect=[None, 0.55]) + cache = await _budget_cache({"virtual_key_spend:test-hash:openai/gpt-4o:7d": 0.55}) result = await _build_model_max_budget_usage( api_key_hash="test-hash", model_max_budget={"openai/gpt-4o": {"budget_limit": 2.0, "time_period": "7d"}}, - user_api_key_cache=mock_user_api_key_cache, + user_api_key_cache=cache, ) + # Seeded only under the configured name, so a provider-stripped probe reads 0.0. assert result["openai/gpt-4o"]["current_spend"] == 0.55 - assert mock_user_api_key_cache.async_get_cache.await_count == 2 def test_list_keys_substring_matching_param_defaults_to_false(): diff --git a/tests/test_litellm/proxy/pass_through_endpoints/test_pass_through_endpoints.py b/tests/test_litellm/proxy/pass_through_endpoints/test_pass_through_endpoints.py index 090acf2dbb0..4c6ba23c88c 100644 --- a/tests/test_litellm/proxy/pass_through_endpoints/test_pass_through_endpoints.py +++ b/tests/test_litellm/proxy/pass_through_endpoints/test_pass_through_endpoints.py @@ -15,9 +15,7 @@ from fastapi import Request, UploadFile from starlette.datastructures import FormData, Headers, QueryParams from starlette.datastructures import UploadFile as StarletteUploadFile -sys.path.insert( - 0, os.path.abspath("../../..") -) # Adds the parent directory to the system path +sys.path.insert(0, os.path.abspath("../../..")) # Adds the parent directory to the system path from litellm.proxy.pass_through_endpoints.pass_through_endpoints import ( DEFAULT_PASS_THROUGH_REQUEST_TIMEOUT_SECONDS, @@ -73,9 +71,7 @@ async def test_build_request_files_from_upload_file(): upload_file = UploadFile(file=file, filename="test.txt", headers=headers) upload_file.read = AsyncMock(return_value=file_content) - result = await HttpPassThroughEndpointHelpers._build_request_files_from_upload_file( - upload_file - ) + result = await HttpPassThroughEndpointHelpers._build_request_files_from_upload_file(upload_file) assert result == ("test.txt", file_content, "text/plain") # Test with Starlette UploadFile @@ -87,9 +83,7 @@ async def test_build_request_files_from_upload_file(): ) starlette_file.read = AsyncMock(return_value=file_content) - result = await HttpPassThroughEndpointHelpers._build_request_files_from_upload_file( - starlette_file - ) + result = await HttpPassThroughEndpointHelpers._build_request_files_from_upload_file(starlette_file) assert result == ("test2.txt", file_content, "text/plain") @@ -275,9 +269,7 @@ async def test_non_streaming_http_request_handler_multipart_with_non_empty_parse """ request = MagicMock(spec=Request) request.method = "POST" - request.headers = Headers( - {"content-type": "multipart/form-data; boundary=------------------------test"} - ) + request.headers = Headers({"content-type": "multipart/form-data; boundary=------------------------test"}) file_content = b"test file content" file = BytesIO(file_content) @@ -316,9 +308,7 @@ 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.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: @@ -329,9 +319,7 @@ async def test_pass_through_request_failure_handler(): # Setup mock for httpx client mock_client = MagicMock() mock_client.client = MagicMock() - mock_client.client.request = AsyncMock( - side_effect=httpx.HTTPError("Request failed") - ) + mock_client.client.request = AsyncMock(side_effect=httpx.HTTPError("Request failed")) mock_get_client.return_value = mock_client # Mock headers for custom headers @@ -364,9 +352,7 @@ async def test_pass_through_request_failure_handler(): # 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 isinstance(call_args["original_exception"], TypeError) # Now expecting TypeError assert "traceback_str" in call_args @@ -410,27 +396,14 @@ def test_is_langfuse_route(): handler = PassThroughEndpointLogging() # Test positive cases - assert ( - handler.is_langfuse_route("http://localhost:4000/langfuse/api/public/traces") - is True - ) - assert ( - handler.is_langfuse_route( - "https://proxy.example.com/langfuse/api/public/sessions" - ) - is True - ) + assert handler.is_langfuse_route("http://localhost:4000/langfuse/api/public/traces") is True + assert handler.is_langfuse_route("https://proxy.example.com/langfuse/api/public/sessions") is True assert handler.is_langfuse_route("/langfuse/api/public/ingestion") is True assert handler.is_langfuse_route("http://localhost:4000/langfuse/") is True # Test negative cases - assert ( - handler.is_langfuse_route("https://api.openai.com/v1/chat/completions") is False - ) - assert ( - handler.is_langfuse_route("http://localhost:4000/anthropic/v1/messages") - is False - ) + assert handler.is_langfuse_route("https://api.openai.com/v1/chat/completions") is False + assert handler.is_langfuse_route("http://localhost:4000/anthropic/v1/messages") is False assert handler.is_langfuse_route("https://example.com/other") is False assert handler.is_langfuse_route("") is False @@ -447,17 +420,9 @@ def test_is_vertex_route_ignores_plain_predict_path_segment(): """ handler = PassThroughEndpointLogging() - assert ( - handler.is_vertex_route( - "https://upstream.example.com/ml/api/v1/time-series-forecast/predict" - ) - is False - ) + assert handler.is_vertex_route("https://upstream.example.com/ml/api/v1/time-series-forecast/predict") is False assert handler.is_vertex_route("https://upstream.example.com/api/v1/search") is False - assert ( - handler.is_vertex_route("https://upstream.example.com/predict/generateContent") - is False - ) + assert handler.is_vertex_route("https://upstream.example.com/predict/generateContent") is False assert ( handler.is_vertex_route( @@ -483,10 +448,7 @@ def test_is_vertex_route_ignores_plain_predict_path_segment(): ) is True ) - assert ( - handler.is_vertex_route("https://discoveryengine.googleapis.com/v1/x:search") - is True - ) + assert handler.is_vertex_route("https://discoveryengine.googleapis.com/v1/x:search") is True assert ( handler.is_vertex_route( @@ -545,9 +507,7 @@ async def test_custom_passthrough_predict_path_logs_via_generic_handler(): mock_vertex_handler.assert_not_called() handler._handle_logging.assert_awaited_once() - logged_object = handler._handle_logging.call_args.kwargs[ - "standard_logging_response_object" - ] + logged_object = handler._handle_logging.call_args.kwargs["standard_logging_response_object"] assert logged_object == {"response": '{"forecast": [1, 2, 3]}'} @@ -600,10 +560,7 @@ async def test_langfuse_passthrough_no_logging(): assert result is None # Verify that the passthrough_logging_payload was still set (this happens before the langfuse check) - assert ( - mock_logging_obj.model_call_details["passthrough_logging_payload"] - == passthrough_logging_payload - ) + assert mock_logging_obj.model_call_details["passthrough_logging_payload"] == passthrough_logging_payload def test_construct_target_url_with_subpath(): @@ -1051,9 +1008,7 @@ async def test_create_pass_through_route_with_cost_per_request(): # Mock the pass_through_request function to capture its call with ( - patch( - "litellm.proxy.pass_through_endpoints.pass_through_endpoints.pass_through_request" - ) as mock_pass_through, + patch("litellm.proxy.pass_through_endpoints.pass_through_endpoints.pass_through_request") as mock_pass_through, patch( "litellm.proxy.pass_through_endpoints.pass_through_endpoints.InitPassThroughEndpointHelpers.is_registered_pass_through_route" ) as mock_is_registered, @@ -1100,10 +1055,7 @@ def test_resolve_pass_through_request_timeout_precedence(): assert resolve_pass_through_request_timeout(endpoint_timeout=800) == 800.0 with patch("litellm.proxy.proxy_server.general_settings", {}): - assert ( - resolve_pass_through_request_timeout() - == DEFAULT_PASS_THROUGH_REQUEST_TIMEOUT_SECONDS - ) + assert resolve_pass_through_request_timeout() == DEFAULT_PASS_THROUGH_REQUEST_TIMEOUT_SECONDS def test_resolve_llm_passthrough_timeout_precedence(): @@ -1135,15 +1087,11 @@ async def test_pass_through_request_uses_resolved_timeout(): with patch( "litellm.proxy.pass_through_endpoints.pass_through_endpoints.get_async_httpx_client" ) as mock_get_client: - mock_proxy_logging.pre_call_hook = AsyncMock( - side_effect=lambda **kwargs: kwargs["data"] - ) + mock_proxy_logging.pre_call_hook = AsyncMock(side_effect=lambda **kwargs: kwargs["data"]) mock_client = MagicMock() mock_client.client = MagicMock() - mock_client.client.request = AsyncMock( - side_effect=httpx.HTTPError("Request failed") - ) + mock_client.client.request = AsyncMock(side_effect=httpx.HTTPError("Request failed")) mock_get_client.return_value = mock_client mock_request = MagicMock(spec=Request) @@ -1181,9 +1129,7 @@ async def test_create_pass_through_route_forwards_timeout(): ) with ( - patch( - "litellm.proxy.pass_through_endpoints.pass_through_endpoints.pass_through_request" - ) as mock_pass_through, + patch("litellm.proxy.pass_through_endpoints.pass_through_endpoints.pass_through_request") as mock_pass_through, patch( "litellm.proxy.pass_through_endpoints.pass_through_endpoints.InitPassThroughEndpointHelpers.is_registered_pass_through_route" ) as mock_is_registered, @@ -1296,9 +1242,7 @@ async def test_pass_through_request_contains_proxy_server_request_in_kwargs(): "litellm.proxy.pass_through_endpoints.pass_through_endpoints.get_response_body" ) as mock_get_response_body: # Setup mock for pre_call_hook and post_call_failure_hook - mock_proxy_logging.pre_call_hook = AsyncMock( - return_value={"test": "data"} - ) + mock_proxy_logging.pre_call_hook = AsyncMock(return_value={"test": "data"}) mock_proxy_logging.post_call_failure_hook = AsyncMock() mock_proxy_logging.post_call_response_headers_hook = AsyncMock( return_value={"x-callback-test": "value"} @@ -1308,9 +1252,7 @@ async def test_pass_through_request_contains_proxy_server_request_in_kwargs(): mock_response = MagicMock() mock_response.status_code = 200 mock_response.headers = {} - mock_response.aread = AsyncMock( - return_value=b'{"success": true}' - ) + mock_response.aread = AsyncMock(return_value=b'{"success": true}') mock_response.text = '{"success": true}' mock_response.raise_for_status = MagicMock() @@ -1330,9 +1272,7 @@ async def test_pass_through_request_contains_proxy_server_request_in_kwargs(): mock_request = MagicMock(spec=Request) mock_request.method = "POST" mock_request.url = "http://test-proxy.com/api/endpoint" - mock_request.body = AsyncMock( - return_value=b'{"message": "test request"}' - ) + mock_request.body = AsyncMock(return_value=b'{"message": "test request"}') mock_request.headers = Headers({}) mock_request.query_params = QueryParams({}) @@ -1411,9 +1351,7 @@ async def test_pass_through_request_streaming_marks_logging_obj_as_stream(): with patch( "litellm.proxy.pass_through_endpoints.pass_through_endpoints.PassThroughStreamingHandler.chunk_processor" ) as mock_chunk_processor: - mock_proxy_logging.pre_call_hook = AsyncMock( - return_value={"model": "claude-3", "stream": True} - ) + mock_proxy_logging.pre_call_hook = AsyncMock(return_value={"model": "claude-3", "stream": True}) mock_proxy_logging.post_call_failure_hook = AsyncMock() mock_proxy_logging.post_call_response_headers_hook = AsyncMock( return_value={"x-callback-test": "value"} @@ -1438,9 +1376,7 @@ async def test_pass_through_request_streaming_marks_logging_obj_as_stream(): mock_request = MagicMock(spec=Request) mock_request.method = "POST" mock_request.url = "http://test-proxy.com/v1/messages" - mock_request.body = AsyncMock( - return_value=b'{"model": "claude-3", "stream": true}' - ) + mock_request.body = AsyncMock(return_value=b'{"model": "claude-3", "stream": true}') mock_request.headers = Headers({}) mock_request.query_params = QueryParams({}) @@ -1456,9 +1392,7 @@ async def test_pass_through_request_streaming_marks_logging_obj_as_stream(): assert async_client.send.call_args.kwargs["stream"] is True mock_chunk_processor.assert_called_once() - logging_obj = mock_chunk_processor.call_args.kwargs[ - "litellm_logging_obj" - ] + logging_obj = mock_chunk_processor.call_args.kwargs["litellm_logging_obj"] assert logging_obj.stream is True assert logging_obj.model_call_details["stream"] is True @@ -1479,9 +1413,7 @@ async def test_pass_through_request_sse_response_marks_logging_obj_as_stream(): with patch( "litellm.proxy.pass_through_endpoints.pass_through_endpoints.PassThroughStreamingHandler.chunk_processor" ) as mock_chunk_processor: - mock_proxy_logging.pre_call_hook = AsyncMock( - return_value={"model": "claude-3"} - ) + mock_proxy_logging.pre_call_hook = AsyncMock(return_value={"model": "claude-3"}) mock_proxy_logging.post_call_failure_hook = AsyncMock() mock_proxy_logging.post_call_response_headers_hook = AsyncMock( return_value={"x-callback-test": "value"} @@ -1521,9 +1453,7 @@ async def test_pass_through_request_sse_response_marks_logging_obj_as_stream(): async_client.send.assert_awaited_once() mock_chunk_processor.assert_called_once() - logging_obj = mock_chunk_processor.call_args.kwargs[ - "litellm_logging_obj" - ] + logging_obj = mock_chunk_processor.call_args.kwargs["litellm_logging_obj"] assert logging_obj.stream is True assert logging_obj.model_call_details["stream"] is True @@ -1550,16 +1480,10 @@ async def test_create_pass_through_endpoint(): ) # Mock the database functions - with patch( - "litellm.proxy.proxy_server.get_config_general_settings" - ) as mock_get_config: - with patch( - "litellm.proxy.proxy_server.update_config_general_settings" - ) as mock_update_config: + with patch("litellm.proxy.proxy_server.get_config_general_settings") as mock_get_config: + with patch("litellm.proxy.proxy_server.update_config_general_settings") as mock_update_config: # Mock existing config (empty list) - mock_get_config.return_value = ConfigFieldInfo( - field_name="pass_through_endpoints", field_value=[] - ) + mock_get_config.return_value = ConfigFieldInfo(field_name="pass_through_endpoints", field_value=[]) # Create test endpoint data test_endpoint = PassThroughGenericEndpoint( @@ -1629,12 +1553,8 @@ async def test_update_pass_through_endpoint(): ) # Mock the database functions - with patch( - "litellm.proxy.proxy_server.get_config_general_settings" - ) as mock_get_config: - with patch( - "litellm.proxy.proxy_server.update_config_general_settings" - ) as mock_update_config: + with patch("litellm.proxy.proxy_server.get_config_general_settings") as mock_get_config: + with patch("litellm.proxy.proxy_server.update_config_general_settings") as mock_update_config: # Create existing endpoint data existing_endpoint_id = "test-endpoint-123" existing_endpoints = [ @@ -1731,18 +1651,14 @@ async def test_create_pass_through_endpoint_auth_true_enforces_allowlist(): registry: dict = {} with ( - patch( - "litellm.proxy.proxy_server.get_config_general_settings" - ) as mock_get_config, + patch("litellm.proxy.proxy_server.get_config_general_settings") as mock_get_config, patch("litellm.proxy.proxy_server.update_config_general_settings"), patch( "litellm.proxy.pass_through_endpoints.pass_through_endpoints._registered_pass_through_routes", registry, ), ): - mock_get_config.return_value = ConfigFieldInfo( - field_name="pass_through_endpoints", field_value=[] - ) + mock_get_config.return_value = ConfigFieldInfo(field_name="pass_through_endpoints", field_value=[]) # auth is not passed -> defaults to True on PassThroughGenericEndpoint endpoint = PassThroughGenericEndpoint( @@ -1757,19 +1673,12 @@ async def test_create_pass_through_endpoint_auth_true_enforces_allowlist(): ) assert any(value.get("auth") is True for value in registry.values()) - assert ( - RouteChecks.is_auth_enforced_pass_through_route( - route="/secure-passthrough", method="POST" - ) - is True - ) + assert RouteChecks.is_auth_enforced_pass_through_route(route="/secure-passthrough", method="POST") is True post_request = MagicMock(spec=Request) post_request.method = "POST" - without_allowlist = UserAPIKeyAuth( - user_id="u", allowed_routes=["llm_api_routes"] - ) + without_allowlist = UserAPIKeyAuth(user_id="u", allowed_routes=["llm_api_routes"]) with pytest.raises(HTTPException) as exc_info: RouteChecks.is_virtual_key_allowed_to_call_route( route="/secure-passthrough", @@ -1826,9 +1735,7 @@ async def test_update_pass_through_endpoint_auth_true_enforces_allowlist(): ] with ( - patch( - "litellm.proxy.proxy_server.get_config_general_settings" - ) as mock_get_config, + patch("litellm.proxy.proxy_server.get_config_general_settings") as mock_get_config, patch("litellm.proxy.proxy_server.update_config_general_settings"), patch( "litellm.proxy.pass_through_endpoints.pass_through_endpoints._registered_pass_through_routes", @@ -1851,19 +1758,12 @@ async def test_update_pass_through_endpoint_auth_true_enforces_allowlist(): user_api_key_dict=MagicMock(spec=UserAPIKeyAuth), ) - assert ( - RouteChecks.is_auth_enforced_pass_through_route( - route="/edited-passthrough", method="POST" - ) - is True - ) + assert RouteChecks.is_auth_enforced_pass_through_route(route="/edited-passthrough", method="POST") is True post_request = MagicMock(spec=Request) post_request.method = "POST" - without_allowlist = UserAPIKeyAuth( - user_id="u", allowed_routes=["llm_api_routes"] - ) + without_allowlist = UserAPIKeyAuth(user_id="u", allowed_routes=["llm_api_routes"]) with pytest.raises(HTTPException) as exc_info: RouteChecks.is_virtual_key_allowed_to_call_route( route="/edited-passthrough", @@ -1905,12 +1805,8 @@ async def test_update_pass_through_endpoint_preserves_auth_false(): ] with ( - patch( - "litellm.proxy.proxy_server.get_config_general_settings" - ) as mock_get_config, - patch( - "litellm.proxy.proxy_server.update_config_general_settings" - ) as mock_update_config, + patch("litellm.proxy.proxy_server.get_config_general_settings") as mock_get_config, + patch("litellm.proxy.proxy_server.update_config_general_settings") as mock_update_config, patch( "litellm.proxy.pass_through_endpoints.pass_through_endpoints._registered_pass_through_routes", registry, @@ -1937,12 +1833,7 @@ async def test_update_pass_through_endpoint_preserves_auth_false(): persisted = mock_update_config.call_args[1]["data"].field_value[0] assert persisted["auth"] is False - assert ( - RouteChecks.is_auth_enforced_pass_through_route( - route="/public-passthrough", method="POST" - ) - is False - ) + assert RouteChecks.is_auth_enforced_pass_through_route(route="/public-passthrough", method="POST") is False @pytest.mark.asyncio @@ -1962,9 +1853,7 @@ async def test_update_pass_through_endpoint_not_found(): ) # Mock the database functions - with patch( - "litellm.proxy.proxy_server.get_config_general_settings" - ) as mock_get_config: + with patch("litellm.proxy.proxy_server.get_config_general_settings") as mock_get_config: # Mock existing config with different endpoint existing_endpoints = [ { @@ -1982,9 +1871,7 @@ async def test_update_pass_through_endpoint_not_found(): ) # Create update data - update_data = PassThroughGenericEndpoint( - path="/test/endpoint", target="http://newapi.com/v2" - ) + update_data = PassThroughGenericEndpoint(path="/test/endpoint", target="http://newapi.com/v2") # Mock user API key dict mock_user_api_key_dict = MagicMock(spec=UserAPIKeyAuth) @@ -2023,12 +1910,8 @@ async def test_delete_pass_through_endpoint(): ) # Mock the database functions - with patch( - "litellm.proxy.proxy_server.get_config_general_settings" - ) as mock_get_config: - with patch( - "litellm.proxy.proxy_server.update_config_general_settings" - ) as mock_update_config: + with patch("litellm.proxy.proxy_server.get_config_general_settings") as mock_get_config: + with patch("litellm.proxy.proxy_server.update_config_general_settings") as mock_update_config: # Create existing endpoint data endpoint_to_delete_id = "test-endpoint-123" other_endpoint_id = "other-endpoint-456" @@ -2106,9 +1989,7 @@ async def test_delete_pass_through_endpoint_not_found(): ) # Mock the database functions - with patch( - "litellm.proxy.proxy_server.get_config_general_settings" - ) as mock_get_config: + with patch("litellm.proxy.proxy_server.get_config_general_settings") as mock_get_config: # Mock existing config with different endpoint existing_endpoints = [ { @@ -2199,14 +2080,8 @@ async def test_get_pass_through_endpoints_includes_config_and_db(): with patch( "litellm.proxy.pass_through_endpoints.pass_through_endpoints._get_pass_through_endpoints_from_config" ) as mock_get_config: - db_objects = [ - PassThroughGenericEndpoint(**ep, is_from_config=False) - for ep in db_endpoints - ] - config_objects = [ - PassThroughGenericEndpoint(**ep, is_from_config=True) - for ep in config_endpoints - ] + db_objects = [PassThroughGenericEndpoint(**ep, is_from_config=False) for ep in db_endpoints] + config_objects = [PassThroughGenericEndpoint(**ep, is_from_config=True) for ep in config_endpoints] mock_get_db.return_value = db_objects mock_get_config.return_value = config_objects @@ -2280,13 +2155,9 @@ async def test_delete_pass_through_endpoint_empty_list(): ) # Mock the database functions - with patch( - "litellm.proxy.proxy_server.get_config_general_settings" - ) as mock_get_config: + with patch("litellm.proxy.proxy_server.get_config_general_settings") as mock_get_config: # Mock empty config - mock_get_config.return_value = ConfigFieldInfo( - field_name="pass_through_endpoints", field_value=None - ) + mock_get_config.return_value = ConfigFieldInfo(field_name="pass_through_endpoints", field_value=None) # Mock user API key dict mock_user_api_key_dict = MagicMock(spec=UserAPIKeyAuth) @@ -2325,9 +2196,7 @@ async def test_pass_through_request_query_params_forwarding(): ) as mock_get_response_body: # Setup mock for pre_call_hook test_body = {"name": "Azure Assistant", "model": "gpt-4o"} - mock_proxy_logging.pre_call_hook = AsyncMock( - return_value=test_body - ) + mock_proxy_logging.pre_call_hook = AsyncMock(return_value=test_body) mock_proxy_logging.post_call_response_headers_hook = AsyncMock( return_value={"x-callback-test": "value"} ) @@ -2336,9 +2205,7 @@ async def test_pass_through_request_query_params_forwarding(): mock_response = MagicMock() mock_response.status_code = 200 mock_response.headers = {"content-type": "application/json"} - mock_response.aread = AsyncMock( - return_value=b'{"id": "asst_123", "object": "assistant"}' - ) + mock_response.aread = AsyncMock(return_value=b'{"id": "asst_123", "object": "assistant"}') mock_response.text = '{"id": "asst_123", "object": "assistant"}' mock_response.raise_for_status = MagicMock() @@ -2360,20 +2227,12 @@ async def test_pass_through_request_query_params_forwarding(): # Create mock request with query parameters (Azure API version) mock_request = MagicMock(spec=Request) mock_request.method = "POST" - mock_request.url = ( - "http://localhost:4000/azure-assistant/openai/assistants" - ) - mock_request.body = AsyncMock( - return_value=json.dumps(test_body).encode() - ) - mock_request.headers = Headers( - {"Content-Type": "application/json"} - ) + mock_request.url = "http://localhost:4000/azure-assistant/openai/assistants" + mock_request.body = AsyncMock(return_value=json.dumps(test_body).encode()) + mock_request.headers = Headers({"Content-Type": "application/json"}) # Create QueryParams with api-version parameter - mock_request.query_params = QueryParams( - [("api-version", "2025-01-01-preview")] - ) + mock_request.query_params = QueryParams([("api-version", "2025-01-01-preview")]) # Create mock user API key dict mock_user_api_key_dict = MagicMock() @@ -2395,9 +2254,7 @@ async def test_pass_through_request_query_params_forwarding(): # The key assertion: query parameters should be preserved and passed to the HTTP handler assert "requested_query_params" in call_kwargs - assert call_kwargs["requested_query_params"] == { - "api-version": "2025-01-01-preview" - } + assert call_kwargs["requested_query_params"] == {"api-version": "2025-01-01-preview"} assert call_kwargs.get("forward_multipart") is False # Verify the target URL is correct @@ -2447,9 +2304,7 @@ async def _run_pass_through_and_capture_wire_url( "PassThroughEndpoint client not found in in_memory_llm_clients_cache; " "get_async_httpx_client may not be caching this provider." ) - cache_dict[cache_key] = SimpleNamespace( - client=httpx.AsyncClient(transport=httpx.MockTransport(transport_handler)) - ) + cache_dict[cache_key] = SimpleNamespace(client=httpx.AsyncClient(transport=httpx.MockTransport(transport_handler))) mock_request = MagicMock(spec=Request) mock_request.method = "GET" @@ -2458,18 +2313,14 @@ async def _run_pass_through_and_capture_wire_url( mock_request.body = AsyncMock(return_value=b"") mock_proxy_logging = MagicMock() - mock_proxy_logging.pre_call_hook = AsyncMock( - side_effect=lambda user_api_key_dict, data, call_type: data - ) + mock_proxy_logging.pre_call_hook = AsyncMock(side_effect=lambda user_api_key_dict, data, call_type: data) mock_proxy_logging.post_call_failure_hook = AsyncMock() mock_proxy_logging.post_call_response_headers_hook = AsyncMock(return_value={}) mock_proxy_logging.get_proxy_hook = MagicMock(return_value=managed_files_hook) try: with ExitStack() as stack: - stack.enter_context( - patch("litellm.proxy.proxy_server.proxy_logging_obj", mock_proxy_logging) - ) + stack.enter_context(patch("litellm.proxy.proxy_server.proxy_logging_obj", mock_proxy_logging)) if managed_files_hook is not None: stack.enter_context( patch( @@ -2637,26 +2488,16 @@ async def test_filter_endpoints_by_team_allowed_routes_with_filter(): # Create test endpoints endpoints = [ - PassThroughGenericEndpoint( - id="endpoint-1", path="/api/allowed1", target="http://example.com/api1" - ), - PassThroughGenericEndpoint( - id="endpoint-2", path="/api/allowed2", target="http://example.com/api2" - ), - PassThroughGenericEndpoint( - id="endpoint-3", path="/api/notallowed", target="http://example.com/api3" - ), + PassThroughGenericEndpoint(id="endpoint-1", path="/api/allowed1", target="http://example.com/api1"), + PassThroughGenericEndpoint(id="endpoint-2", path="/api/allowed2", target="http://example.com/api2"), + PassThroughGenericEndpoint(id="endpoint-3", path="/api/notallowed", target="http://example.com/api3"), ] # Mock prisma client mock_prisma_client = MagicMock() mock_team = MagicMock() - mock_team.metadata = { - "allowed_passthrough_routes": ["/api/allowed1", "/api/allowed2"] - } - mock_prisma_client.db.litellm_teamtable.find_unique = AsyncMock( - return_value=mock_team - ) + mock_team.metadata = {"allowed_passthrough_routes": ["/api/allowed1", "/api/allowed2"]} + mock_prisma_client.db.litellm_teamtable.find_unique = AsyncMock(return_value=mock_team) # Call the function result = await _filter_endpoints_by_team_allowed_routes( @@ -2671,9 +2512,7 @@ async def test_filter_endpoints_by_team_allowed_routes_with_filter(): assert result[1].path == "/api/allowed2" # Verify database call - mock_prisma_client.db.litellm_teamtable.find_unique.assert_called_once_with( - where={"team_id": "test-team-123"} - ) + mock_prisma_client.db.litellm_teamtable.find_unique.assert_called_once_with(where={"team_id": "test-team-123"}) @pytest.mark.asyncio @@ -2691,9 +2530,7 @@ async def test_filter_endpoints_by_team_allowed_routes_team_not_found(): # Create test endpoints endpoints = [ - PassThroughGenericEndpoint( - id="endpoint-1", path="/api/test", target="http://example.com/api" - ), + PassThroughGenericEndpoint(id="endpoint-1", path="/api/test", target="http://example.com/api"), ] # Mock prisma client to return None (team not found) @@ -2726,21 +2563,15 @@ async def test_filter_endpoints_by_team_allowed_routes_no_metadata(): # Create test endpoints endpoints = [ - PassThroughGenericEndpoint( - id="endpoint-1", path="/api/test1", target="http://example.com/api1" - ), - PassThroughGenericEndpoint( - id="endpoint-2", path="/api/test2", target="http://example.com/api2" - ), + PassThroughGenericEndpoint(id="endpoint-1", path="/api/test1", target="http://example.com/api1"), + PassThroughGenericEndpoint(id="endpoint-2", path="/api/test2", target="http://example.com/api2"), ] # Mock prisma client with team that has None metadata mock_prisma_client = MagicMock() mock_team = MagicMock() mock_team.metadata = None - mock_prisma_client.db.litellm_teamtable.find_unique = AsyncMock( - return_value=mock_team - ) + mock_prisma_client.db.litellm_teamtable.find_unique = AsyncMock(return_value=mock_team) # Call the function result = await _filter_endpoints_by_team_allowed_routes( @@ -2768,21 +2599,15 @@ async def test_filter_endpoints_by_team_allowed_routes_no_allowed_routes_key(): # Create test endpoints endpoints = [ - PassThroughGenericEndpoint( - id="endpoint-1", path="/api/test1", target="http://example.com/api1" - ), - PassThroughGenericEndpoint( - id="endpoint-2", path="/api/test2", target="http://example.com/api2" - ), + PassThroughGenericEndpoint(id="endpoint-1", path="/api/test1", target="http://example.com/api1"), + PassThroughGenericEndpoint(id="endpoint-2", path="/api/test2", target="http://example.com/api2"), ] # Mock prisma client with team that has metadata but no allowed_passthrough_routes mock_prisma_client = MagicMock() mock_team = MagicMock() mock_team.metadata = {"some_other_key": "some_value"} - mock_prisma_client.db.litellm_teamtable.find_unique = AsyncMock( - return_value=mock_team - ) + mock_prisma_client.db.litellm_teamtable.find_unique = AsyncMock(return_value=mock_team) # Call the function result = await _filter_endpoints_by_team_allowed_routes( @@ -2810,21 +2635,15 @@ async def test_filter_endpoints_by_team_allowed_routes_empty_allowed_list(): # Create test endpoints endpoints = [ - PassThroughGenericEndpoint( - id="endpoint-1", path="/api/test1", target="http://example.com/api1" - ), - PassThroughGenericEndpoint( - id="endpoint-2", path="/api/test2", target="http://example.com/api2" - ), + PassThroughGenericEndpoint(id="endpoint-1", path="/api/test1", target="http://example.com/api1"), + PassThroughGenericEndpoint(id="endpoint-2", path="/api/test2", target="http://example.com/api2"), ] # Mock prisma client with team that has empty allowed_passthrough_routes mock_prisma_client = MagicMock() mock_team = MagicMock() mock_team.metadata = {"allowed_passthrough_routes": []} - mock_prisma_client.db.litellm_teamtable.find_unique = AsyncMock( - return_value=mock_team - ) + mock_prisma_client.db.litellm_teamtable.find_unique = AsyncMock(return_value=mock_team) # Call the function result = await _filter_endpoints_by_team_allowed_routes( @@ -2850,29 +2669,21 @@ async def test_filter_endpoints_by_team_allowed_routes_partial_match(): # Create test endpoints endpoints = [ - PassThroughGenericEndpoint( - id="endpoint-1", path="/api/openai", target="http://example.com/openai" - ), + PassThroughGenericEndpoint(id="endpoint-1", path="/api/openai", target="http://example.com/openai"), PassThroughGenericEndpoint( id="endpoint-2", path="/api/anthropic", target="http://example.com/anthropic", ), - PassThroughGenericEndpoint( - id="endpoint-3", path="/api/azure", target="http://example.com/azure" - ), - PassThroughGenericEndpoint( - id="endpoint-4", path="/api/cohere", target="http://example.com/cohere" - ), + PassThroughGenericEndpoint(id="endpoint-3", path="/api/azure", target="http://example.com/azure"), + PassThroughGenericEndpoint(id="endpoint-4", path="/api/cohere", target="http://example.com/cohere"), ] # Mock prisma client with team that allows only 2 routes mock_prisma_client = MagicMock() mock_team = MagicMock() mock_team.metadata = {"allowed_passthrough_routes": ["/api/openai", "/api/azure"]} - mock_prisma_client.db.litellm_teamtable.find_unique = AsyncMock( - return_value=mock_team - ) + mock_prisma_client.db.litellm_teamtable.find_unique = AsyncMock(return_value=mock_team) # Call the function result = await _filter_endpoints_by_team_allowed_routes( @@ -2904,9 +2715,7 @@ async def test_bedrock_router_passthrough_metadata_initialization(): ) # Mock ProxyBaseLLMRequestProcessing to verify it's used - with patch( - "litellm.proxy.common_request_processing.ProxyBaseLLMRequestProcessing" - ) as mock_processing_class: + with patch("litellm.proxy.common_request_processing.ProxyBaseLLMRequestProcessing") as mock_processing_class: # Setup mock instance mock_processor = MagicMock() mock_processing_class.return_value = mock_processor @@ -2914,12 +2723,8 @@ async def test_bedrock_router_passthrough_metadata_initialization(): # Mock successful response mock_response = MagicMock() mock_response.status_code = 200 - mock_response.aread = AsyncMock( - return_value=b'{"content": [{"text": "Hello"}]}' - ) - mock_processor.base_passthrough_process_llm_request = AsyncMock( - return_value=mock_response - ) + mock_response.aread = AsyncMock(return_value=b'{"content": [{"text": "Hello"}]}') + mock_processor.base_passthrough_process_llm_request = AsyncMock(return_value=mock_response) # Create mock request with headers mock_request = MagicMock(spec=Request) @@ -2986,18 +2791,10 @@ async def test_bedrock_router_passthrough_metadata_initialization(): call_kwargs = mock_processor.base_passthrough_process_llm_request.call_args[1] # These are the critical parameters that ensure metadata is properly initialized: - assert ( - call_kwargs["request"] == mock_request - ), "Request must be passed for header extraction" - assert ( - call_kwargs["user_api_key_dict"] == mock_user_api_key_dict - ), "User API key dict needed for metadata" - assert ( - call_kwargs["proxy_logging_obj"] == mock_proxy_logging - ), "Logging obj needed for hooks" - assert ( - call_kwargs["llm_router"] == mock_router - ), "Router needed for model routing" + assert call_kwargs["request"] == mock_request, "Request must be passed for header extraction" + assert call_kwargs["user_api_key_dict"] == mock_user_api_key_dict, "User API key dict needed for metadata" + assert call_kwargs["proxy_logging_obj"] == mock_proxy_logging, "Logging obj needed for hooks" + assert call_kwargs["llm_router"] == mock_router, "Router needed for model routing" assert call_kwargs["model"] == "my-bedrock-model", "Model name must be passed" # Verify response was returned @@ -3060,18 +2857,12 @@ async def test_add_litellm_data_to_request_adds_headers_to_metadata(): # Bedrock passthrough uses litellm_metadata to prevent key-level # tags from leaking into the provider payload (GH#30629). assert "litellm_metadata" in result, "litellm_metadata should be present in result" - assert ( - "headers" in result["litellm_metadata"] - ), "headers should be present in litellm_metadata" - assert isinstance( - result["litellm_metadata"]["headers"], dict - ), "headers should be a dictionary" + assert "headers" in result["litellm_metadata"], "headers should be present in litellm_metadata" + assert isinstance(result["litellm_metadata"]["headers"], dict), "headers should be a dictionary" # Verify specific headers are accessible (important for guardrails) headers = result["litellm_metadata"]["headers"] - assert ( - "user-agent" in headers or "User-Agent" in headers - ), "User-Agent header should be accessible in metadata" + assert "user-agent" in headers or "User-Agent" in headers, "User-Agent header should be accessible in metadata" # Also verify proxy_server_request has headers (original location) assert "proxy_server_request" in result @@ -3106,9 +2897,7 @@ async def test_create_pass_through_route_custom_body_url_target(): ) with ( - patch( - "litellm.proxy.pass_through_endpoints.pass_through_endpoints.pass_through_request" - ) as mock_pass_through, + patch("litellm.proxy.pass_through_endpoints.pass_through_endpoints.pass_through_request") as mock_pass_through, patch( "litellm.proxy.pass_through_endpoints.pass_through_endpoints.InitPassThroughEndpointHelpers.is_registered_pass_through_route" ) as mock_is_registered, @@ -3145,9 +2934,7 @@ async def test_create_pass_through_route_custom_body_url_target(): "retrievalQuery": {"text": "What is in the knowledge base?"}, } - setattr( - mock_request.state, LITELLM_PASS_THROUGH_CUSTOM_BODY_STATE_KEY, bedrock_body - ) + setattr(mock_request.state, LITELLM_PASS_THROUGH_CUSTOM_BODY_STATE_KEY, bedrock_body) await endpoint_func( request=mock_request, @@ -3185,9 +2972,7 @@ async def test_pass_through_request_non_streaming_uses_content_for_state_raw_bod mock_request.headers = Headers({"Content-Type": "application/json"}) mock_request.state = SimpleNamespace() setattr(mock_request.state, LITELLM_PASS_THROUGH_RAW_BODY_STATE_KEY, raw_signed) - mock_request.body = AsyncMock( - return_value=json.dumps(parsed_from_wire).encode("utf-8") - ) + mock_request.body = AsyncMock(return_value=json.dumps(parsed_from_wire).encode("utf-8")) mock_user = MagicMock() mock_user.api_key = "sk-test" @@ -3260,9 +3045,7 @@ async def test_pass_through_request_streaming_uses_content_for_state_raw_body(): mock_request.headers = Headers({"Content-Type": "application/json"}) mock_request.state = SimpleNamespace() setattr(mock_request.state, LITELLM_PASS_THROUGH_RAW_BODY_STATE_KEY, raw_signed) - mock_request.body = AsyncMock( - return_value=json.dumps(parsed_from_wire).encode("utf-8") - ) + mock_request.body = AsyncMock(return_value=json.dumps(parsed_from_wire).encode("utf-8")) mock_user = MagicMock() mock_user.api_key = "sk-test" @@ -3336,9 +3119,7 @@ async def test_create_pass_through_route_no_custom_body_falls_back(): ) with ( - patch( - "litellm.proxy.pass_through_endpoints.pass_through_endpoints.pass_through_request" - ) as mock_pass_through, + patch("litellm.proxy.pass_through_endpoints.pass_through_endpoints.pass_through_request") as mock_pass_through, patch( "litellm.proxy.pass_through_endpoints.pass_through_endpoints.InitPassThroughEndpointHelpers.is_registered_pass_through_route" ) as mock_is_registered, @@ -3407,32 +3188,12 @@ def test_is_registered_pass_through_route_with_custom_root(): } with patch("litellm.proxy.utils.get_server_root_path", return_value="/proxy"): - assert ( - InitPassThroughEndpointHelpers.is_registered_pass_through_route( - "/proxy/api/endpoint" - ) - is True - ) - assert ( - InitPassThroughEndpointHelpers.is_registered_pass_through_route( - "/api/endpoint" - ) - is True - ) + assert InitPassThroughEndpointHelpers.is_registered_pass_through_route("/proxy/api/endpoint") is True + assert InitPassThroughEndpointHelpers.is_registered_pass_through_route("/api/endpoint") is True with patch("litellm.proxy.utils.get_server_root_path", return_value="/"): - assert ( - InitPassThroughEndpointHelpers.is_registered_pass_through_route( - "/api/endpoint" - ) - is True - ) - assert ( - InitPassThroughEndpointHelpers.is_registered_pass_through_route( - "/proxy/api/endpoint" - ) - is False - ) + assert InitPassThroughEndpointHelpers.is_registered_pass_through_route("/api/endpoint") is True + assert InitPassThroughEndpointHelpers.is_registered_pass_through_route("/proxy/api/endpoint") is False # Clean up _registered_pass_through_routes.clear() @@ -3464,24 +3225,18 @@ def test_get_registered_pass_through_route_with_custom_root(): with patch("litellm.proxy.utils.get_server_root_path", return_value="/litellm"): # Prefixed incoming route - result = InitPassThroughEndpointHelpers.get_registered_pass_through_route( - "/litellm/chat/completions" - ) + result = InitPassThroughEndpointHelpers.get_registered_pass_through_route("/litellm/chat/completions") assert result is not None assert result["target"] == "http://api.example.com/v1/chat/completions" assert result["headers"]["Authorization"] == "Bearer token123" # Bare incoming route (get_request_route convention) - result = InitPassThroughEndpointHelpers.get_registered_pass_through_route( - "/chat/completions" - ) + result = InitPassThroughEndpointHelpers.get_registered_pass_through_route("/chat/completions") assert result is not None assert result["target"] == "http://api.example.com/v1/chat/completions" with patch("litellm.proxy.utils.get_server_root_path", return_value="/"): - result = InitPassThroughEndpointHelpers.get_registered_pass_through_route( - "/chat/completions" - ) + result = InitPassThroughEndpointHelpers.get_registered_pass_through_route("/chat/completions") assert result is not None assert result["target"] == "http://api.example.com/v1/chat/completions" @@ -3535,12 +3290,7 @@ def test_db_registered_pass_through_route_bare_path_convention( "litellm.proxy.utils.get_server_root_path", return_value=server_root_path, ): - assert ( - InitPassThroughEndpointHelpers.is_registered_pass_through_route( - incoming_route - ) - is should_match - ) + assert InitPassThroughEndpointHelpers.is_registered_pass_through_route(incoming_route) is should_match _registered_pass_through_routes.clear() @@ -3559,25 +3309,13 @@ def test_mapped_pass_through_routes_with_server_root_path(): with patch("litellm.proxy.utils.get_server_root_path", return_value="/litellm"): # prefixed route should match mapped routes like /vertex_ai assert ( - InitPassThroughEndpointHelpers.is_registered_pass_through_route( - "/litellm/vertex_ai/v1/projects/foo" - ) - is True - ) - assert ( - InitPassThroughEndpointHelpers.is_registered_pass_through_route( - "/litellm/bedrock/model/invoke" - ) + InitPassThroughEndpointHelpers.is_registered_pass_through_route("/litellm/vertex_ai/v1/projects/foo") is True ) + assert InitPassThroughEndpointHelpers.is_registered_pass_through_route("/litellm/bedrock/model/invoke") is True # bare route without prefix should not match when root is set - assert ( - InitPassThroughEndpointHelpers.is_registered_pass_through_route( - "/vertex_ai/v1/projects/foo" - ) - is False - ) + assert InitPassThroughEndpointHelpers.is_registered_pass_through_route("/vertex_ai/v1/projects/foo") is False @pytest.mark.asyncio @@ -3594,24 +3332,18 @@ async def test_multipart_passthrough_preserves_boundary(): mock_response = MagicMock() mock_response.status_code = 200 mock_response.headers = httpx.Headers({"content-type": "application/json"}) - mock_response.aread = AsyncMock( - return_value=b'{"filename": "test.txt", "size": 17}' - ) + mock_response.aread = AsyncMock(return_value=b'{"filename": "test.txt", "size": 17}') mock_response.text = '{"filename": "test.txt", "size": 17}' async def mock_httpx_request(method, url, **kwargs): # Verify that files parameter is passed (not json) assert "files" in kwargs, "Files should be passed for multipart requests" - file_parts = [ - value for name, value in kwargs["files"] if name == "file" - ] + file_parts = [value for name, value in kwargs["files"] if name == "file"] assert len(file_parts) == 1, "File field should be in files" # Verify content-type is NOT in headers (httpx will set it with correct boundary) headers = kwargs.get("headers", {}) - assert ( - "content-type" not in headers - ), "content-type should be removed for multipart" + assert "content-type" not in headers, "content-type should be removed for multipart" filename, content, content_type = file_parts[0] assert filename == "test.txt" @@ -3684,9 +3416,7 @@ def test_get_response_headers_strips_server_and_date(): "connection", "keep-alive", ): - assert ( - stripped not in lowered_keys - ), f"{stripped!r} must not be forwarded by passthrough" + assert stripped not in lowered_keys, f"{stripped!r} must not be forwarded by passthrough" # Application/business headers must still pass through. lowered = {k.lower(): v for k, v in result.items()} @@ -3724,9 +3454,7 @@ class TestStaleRouteCleanupOnReload: ) stack.enter_context(patch("litellm.proxy.proxy_server.premium_user", True)) mock_set_env = stack.enter_context( - patch( - "litellm.proxy.pass_through_endpoints.pass_through_endpoints.set_env_variables_in_header" - ) + patch("litellm.proxy.pass_through_endpoints.pass_through_endpoints.set_env_variables_in_header") ) mock_set_env.return_value = {} return stack @@ -3771,14 +3499,10 @@ class TestStaleRouteCleanupOnReload: so the registry would hold both paths instead of only ``/b``. """ with self._patches(): - await initialize_pass_through_endpoints( - [{"path": "/a", "target": "http://example.com"}] - ) + await initialize_pass_through_endpoints([{"path": "/a", "target": "http://example.com"}]) assert self._paths_in_registry() == ["/a"] - await initialize_pass_through_endpoints( - [{"path": "/b", "target": "http://example.com"}] - ) + await initialize_pass_through_endpoints([{"path": "/b", "target": "http://example.com"}]) assert self._paths_in_registry() == ["/b"] @@ -3801,12 +3525,8 @@ class TestStaleRouteCleanupOnReload: ] ) - assert InitPassThroughEndpointHelpers.is_registered_pass_through_route( - "/live-passthrough" - ) - assert InitPassThroughEndpointHelpers.is_registered_pass_through_route( - "/live-passthrough/some/subpath" - ) + assert InitPassThroughEndpointHelpers.is_registered_pass_through_route("/live-passthrough") + assert InitPassThroughEndpointHelpers.is_registered_pass_through_route("/live-passthrough/some/subpath") # Regression (LIT-3538): a pre-call guardrail block on a passthrough endpoint @@ -3907,40 +3627,26 @@ async def _drive_pass_through_block(raised_exception): 400, ), ( - _FastAPIHTTPException( - status_code=400, detail={"error": "Violated moderation policy"} - ), + _FastAPIHTTPException(status_code=400, detail={"error": "Violated moderation policy"}), 400, ), ], ) -async def test_pre_call_guardrail_block_logs_warning_not_exception( - guardrail_exception, expected_code -): +async def test_pre_call_guardrail_block_logs_warning_not_exception(guardrail_exception, expected_code): status_code, logger = await _drive_pass_through_block(guardrail_exception) assert int(status_code) == expected_code - assert ( - logger.exception.call_count == 0 - ), "guardrail block must not be logged as an ERROR with a traceback" - assert ( - logger.warning.call_count == 1 - ), "guardrail block must be logged once at WARNING" + assert logger.exception.call_count == 0, "guardrail block must not be logged as an ERROR with a traceback" + assert logger.warning.call_count == 1, "guardrail block must be logged once at WARNING" @pytest.mark.asyncio async def test_non_guardrail_exception_still_logs_with_traceback(): - status_code, logger = await _drive_pass_through_block( - RuntimeError("upstream connection reset") - ) + status_code, logger = await _drive_pass_through_block(RuntimeError("upstream connection reset")) assert int(status_code) == 500 - assert ( - logger.exception.call_count == 1 - ), "a genuine failure must still be logged via verbose_proxy_logger.exception" - assert ( - logger.warning.call_count == 0 - ), "a genuine failure must not be downgraded to WARNING" + assert logger.exception.call_count == 1, "a genuine failure must still be logged via verbose_proxy_logger.exception" + assert logger.warning.call_count == 0, "a genuine failure must not be downgraded to WARNING" # Regression: generic config-based passthrough (`pass_through_request`) used to @@ -3979,9 +3685,7 @@ async def test_pass_through_request_non_streaming_upstream_error_returned_unchan ) as mock_success_handler: mock_proxy_logging.pre_call_hook = AsyncMock(return_value={}) mock_proxy_logging.post_call_failure_hook = AsyncMock() - mock_proxy_logging.post_call_response_headers_hook = AsyncMock( - return_value=None - ) + mock_proxy_logging.post_call_response_headers_hook = AsyncMock(return_value=None) mock_processing.get_custom_headers.return_value = {} mock_success_handler.return_value = None @@ -4067,9 +3771,7 @@ async def test_pass_through_request_upstream_error_failure_hook_exception_is_swa mock_proxy_logging.post_call_failure_hook = AsyncMock( side_effect=RuntimeError("alerting integration misconfigured") ) - mock_proxy_logging.post_call_response_headers_hook = AsyncMock( - return_value=None - ) + mock_proxy_logging.post_call_response_headers_hook = AsyncMock(return_value=None) mock_processing.get_custom_headers.return_value = {} mock_success_handler.return_value = None @@ -4119,9 +3821,7 @@ async def test_pass_through_request_streaming_upstream_error_returned_unchanged( ) as mock_success_handler: mock_proxy_logging.pre_call_hook = AsyncMock(return_value={}) mock_proxy_logging.post_call_failure_hook = AsyncMock() - mock_proxy_logging.post_call_response_headers_hook = AsyncMock( - return_value=None - ) + mock_proxy_logging.post_call_response_headers_hook = AsyncMock(return_value=None) mock_success_handler.return_value = None async_client = MagicMock() @@ -4149,10 +3849,7 @@ async def test_pass_through_request_streaming_upstream_error_returned_unchanged( streamed_chunks = [chunk async for chunk in response.body_iterator] await asyncio.sleep(0) - streamed_bytes = b"".join( - chunk if isinstance(chunk, bytes) else chunk.encode("utf-8") - for chunk in streamed_chunks - ) + streamed_bytes = b"".join(chunk if isinstance(chunk, bytes) else chunk.encode("utf-8") for chunk in streamed_chunks) assert streamed_bytes == upstream_content assert json.loads(streamed_bytes) == _UPSTREAM_ERROR_BODY @@ -4198,9 +3895,7 @@ async def test_pass_through_request_non_streaming_success_unchanged(): ) as mock_success_handler: mock_proxy_logging.pre_call_hook = AsyncMock(return_value={}) mock_proxy_logging.post_call_failure_hook = AsyncMock() - mock_proxy_logging.post_call_response_headers_hook = AsyncMock( - return_value=None - ) + mock_proxy_logging.post_call_response_headers_hook = AsyncMock(return_value=None) mock_processing.get_custom_headers.return_value = {} mock_success_handler.return_value = None @@ -4244,9 +3939,7 @@ async def test_pass_through_request_internal_failure_still_raises_proxy_exceptio from litellm.proxy._types import ProxyException with patch("litellm.proxy.proxy_server.proxy_logging_obj") as mock_proxy_logging: - mock_proxy_logging.pre_call_hook = AsyncMock( - side_effect=RuntimeError("auth backend unavailable") - ) + mock_proxy_logging.pre_call_hook = AsyncMock(side_effect=RuntimeError("auth backend unavailable")) mock_proxy_logging.post_call_failure_hook = AsyncMock() mock_request = MagicMock(spec=Request) @@ -4335,9 +4028,7 @@ def _inject_fake_passthrough_client(transport, timeout): def _enter_relay_logging_mocks(stack, parsed_body): from litellm.litellm_core_utils.logging_worker import GLOBAL_LOGGING_WORKER - mock_proxy_logging = stack.enter_context( - patch("litellm.proxy.proxy_server.proxy_logging_obj") - ) + mock_proxy_logging = stack.enter_context(patch("litellm.proxy.proxy_server.proxy_logging_obj")) mock_proxy_logging.pre_call_hook = AsyncMock(return_value=parsed_body) mock_proxy_logging.post_call_failure_hook = AsyncMock() mock_proxy_logging.post_call_response_headers_hook = AsyncMock(return_value=None) @@ -4347,11 +4038,7 @@ def _enter_relay_logging_mocks(stack, parsed_body): ) ) mock_success_handler.return_value = None - stack.enter_context( - patch.object( - GLOBAL_LOGGING_WORKER, "ensure_initialized_and_enqueue", new=MagicMock() - ) - ) + stack.enter_context(patch.object(GLOBAL_LOGGING_WORKER, "ensure_initialized_and_enqueue", new=MagicMock())) return mock_proxy_logging, mock_success_handler @@ -4437,10 +4124,7 @@ async def test_pass_through_request_relays_non_json_body_without_buffering(): mock_success_handler.assert_called_once() success_kwargs = mock_success_handler.call_args.kwargs assert success_kwargs["response_body"] is None - assert ( - success_kwargs["url_route"] - == "http://upstream.test/v1/messages/batches/b1/results" - ) + assert success_kwargs["url_route"] == "http://upstream.test/v1/messages/batches/b1/results" finally: cleanup() await fake_client.aclose() @@ -4518,9 +4202,7 @@ async def test_pass_through_request_upstream_error_body_stays_buffered(): ) try: with ExitStack() as stack: - mock_proxy_logging, mock_success_handler = _enter_relay_logging_mocks( - stack, {} - ) + mock_proxy_logging, mock_success_handler = _enter_relay_logging_mocks(stack, {}) response = await pass_through_request( request=_relay_client_request(), @@ -4590,18 +4272,11 @@ async def test_pass_through_relay_client_disconnect_logs_partial_relay_warning(c partial_relay_warnings = [ record.getMessage() for record in caplog.records - if record.levelno == logging.WARNING - and _PARTIAL_RELAY_WARNING_MARKER in record.getMessage() + if record.levelno == logging.WARNING and _PARTIAL_RELAY_WARNING_MARKER in record.getMessage() ] assert len(partial_relay_warnings) == 1 - assert ( - "http://upstream.test/v1/messages/batches/b1/results" - in partial_relay_warnings[0] - ) - assert ( - f"{len(first_chunk)} bytes were sent to the client" - in partial_relay_warnings[0] - ) + assert "http://upstream.test/v1/messages/batches/b1/results" in partial_relay_warnings[0] + assert f"{len(first_chunk)} bytes were sent to the client" in partial_relay_warnings[0] assert upstream_stream.closed is True mock_success_handler.assert_called_once() @@ -4648,10 +4323,7 @@ async def test_pass_through_relay_full_consumption_logs_no_partial_relay_warning relayed = [chunk async for chunk in response.body_iterator] assert b"".join(relayed) == b"".join(upstream_chunks) - assert not any( - _PARTIAL_RELAY_WARNING_MARKER in record.getMessage() - for record in caplog.records - ) + assert not any(_PARTIAL_RELAY_WARNING_MARKER in record.getMessage() for record in caplog.records) mock_success_handler.assert_called_once() finally: cleanup() @@ -4689,9 +4361,7 @@ def _enter_upstream_usage_mocks(stack, parsed_body): logging worker would have run so the test can await them.""" from litellm.litellm_core_utils.logging_worker import GLOBAL_LOGGING_WORKER - mock_proxy_logging = stack.enter_context( - patch("litellm.proxy.proxy_server.proxy_logging_obj") - ) + mock_proxy_logging = stack.enter_context(patch("litellm.proxy.proxy_server.proxy_logging_obj")) mock_proxy_logging.pre_call_hook = AsyncMock(return_value=parsed_body) mock_proxy_logging.post_call_failure_hook = AsyncMock() mock_proxy_logging.post_call_response_headers_hook = AsyncMock(return_value=None) @@ -4708,9 +4378,7 @@ def _enter_upstream_usage_mocks(stack, parsed_body): return mock_proxy_logging, enqueued -async def _run_upstream_reporting_passthrough( - upstream_headers, status_code=200, cost_per_request=None -): +async def _run_upstream_reporting_passthrough(upstream_headers, status_code=200, cost_per_request=None): """Drive a generic pass-through against an upstream that reports its own cost/usage. Returns (recorded standard logging payloads, proxy logging mock).""" from litellm.proxy._types import UserAPIKeyAuth @@ -4731,9 +4399,7 @@ async def _run_upstream_reporting_passthrough( request=_relay_client_request(method="POST"), target="http://internal-api.test/v1/summarize", custom_headers={}, - user_api_key_dict=UserAPIKeyAuth( - api_key="sk-upstream-usage", team_id="team-fil" - ), + user_api_key_dict=UserAPIKeyAuth(api_key="sk-upstream-usage", team_id="team-fil"), cost_per_request=cost_per_request, ) for coroutine in enqueued: @@ -4802,23 +4468,17 @@ async def test_passthrough_records_upstream_reported_cost_on_error_response(): ) mock_proxy_logging.post_call_failure_hook.assert_awaited_once() - request_data = mock_proxy_logging.post_call_failure_hook.await_args.kwargs[ - "request_data" - ] + request_data = mock_proxy_logging.post_call_failure_hook.await_args.kwargs["request_data"] assert request_data["response_cost"] == 0.00021 assert request_data["combined_usage_object"] == litellm.Usage(total_tokens=930) @pytest.mark.asyncio async def test_passthrough_error_response_without_usage_headers_records_no_spend(): - _, mock_proxy_logging = await _run_upstream_reporting_passthrough( - {}, status_code=500 - ) + _, mock_proxy_logging = await _run_upstream_reporting_passthrough({}, status_code=500) mock_proxy_logging.post_call_failure_hook.assert_awaited_once() - request_data = mock_proxy_logging.post_call_failure_hook.await_args.kwargs[ - "request_data" - ] + request_data = mock_proxy_logging.post_call_failure_hook.await_args.kwargs["request_data"] assert "combined_usage_object" not in request_data @@ -4853,14 +4513,10 @@ async def test_streaming_passthrough_records_cost_and_tokens_reported_by_upstrea request=_relay_client_request(method="POST"), target="http://internal-api.test/v1/summarize", custom_headers={}, - user_api_key_dict=UserAPIKeyAuth( - api_key="sk-upstream-usage", team_id="team-fil" - ), + user_api_key_dict=UserAPIKeyAuth(api_key="sk-upstream-usage", team_id="team-fil"), ) assert isinstance(response, StreamingResponse) - assert [chunk async for chunk in response.body_iterator] == [ - b'data: {"delta": "hi"}\n\n' - ] + assert [chunk async for chunk in response.body_iterator] == [b'data: {"delta": "hi"}\n\n'] for coroutine in enqueued: await coroutine finally: @@ -4968,9 +4624,7 @@ async def test_websocket_passthrough_forwards_non_ascii_first_frame(): "litellm.proxy.pass_through_endpoints.pass_through_endpoints.connect", return_value=FakeUpstreamConnect(upstream_ws), ), - patch( - "litellm.proxy.pass_through_endpoints.pass_through_endpoints.GLOBAL_LOGGING_WORKER" - ) as mock_worker, + patch("litellm.proxy.pass_through_endpoints.pass_through_endpoints.GLOBAL_LOGGING_WORKER") as mock_worker, ): mock_proxy_logging.pre_call_hook = AsyncMock(return_value={}) mock_proxy_logging.post_call_success_hook = AsyncMock() @@ -5052,9 +4706,7 @@ def _patched_websocket_passthrough_environment(upstream_ws): "litellm.proxy.pass_through_endpoints.pass_through_endpoints.connect", return_value=FakeUpstreamConnect(upstream_ws), ), - patch( - "litellm.proxy.pass_through_endpoints.pass_through_endpoints.GLOBAL_LOGGING_WORKER" - ) as mock_worker, + patch("litellm.proxy.pass_through_endpoints.pass_through_endpoints.GLOBAL_LOGGING_WORKER") as mock_worker, ): mock_proxy_logging.pre_call_hook = AsyncMock(return_value={}) mock_proxy_logging.post_call_success_hook = AsyncMock() @@ -5199,9 +4851,7 @@ async def test_websocket_passthrough_does_not_relay_unsendable_upstream_close(rc "abnormal": Close(1006, "connection died"), "no_status": Close(1005, ""), }[rcvd_close] - upstream_ws = ClosingUpstreamWebSocket( - ConnectionClosedError(rcvd=rcvd, sent=None, rcvd_then_sent=None) - ) + upstream_ws = ClosingUpstreamWebSocket(ConnectionClosedError(rcvd=rcvd, sent=None, rcvd_then_sent=None)) websocket = _client_websocket(_pending_receive) with _patched_websocket_passthrough_environment(upstream_ws): @@ -5276,14 +4926,15 @@ async def test_websocket_passthrough_does_not_close_twice_when_success_logging_f def _passthrough_kwargs_for_reservation( - user_api_key_dict: UserAPIKeyAuth, parsed_body: Optional[dict] = None + user_api_key_dict: UserAPIKeyAuth, + parsed_body: Optional[dict] = None, + user_defined_route: bool = False, ) -> dict: mock_request = MagicMock(spec=Request) mock_request.method = "POST" - mock_request.url = ( - "http://0.0.0.0:4000/gemini/v1beta/models/gemini-2.5-flash:generateContent" - ) + mock_request.url = "http://0.0.0.0:4000/gemini/v1beta/models/gemini-2.5-flash:generateContent" mock_request.headers = Headers({}) + mock_request.scope = {"endpoint": _marked_pass_through_endpoint()} if user_defined_route else {} return HttpPassThroughEndpointHelpers._init_kwargs_for_pass_through_endpoint( request=mock_request, @@ -5352,10 +5003,7 @@ async def test_passthrough_success_reconciles_budget_reservation(): reservation = user_api_key_dict.budget_reservation kwargs = _passthrough_kwargs_for_reservation(user_api_key_dict) - assert ( - kwargs["litellm_params"]["metadata"]["user_api_key_budget_reservation"] - is reservation - ) + assert kwargs["litellm_params"]["metadata"]["user_api_key_budget_reservation"] is reservation increment_spend_counters = await _track_cost_for_passthrough_kwargs(kwargs) @@ -5381,9 +5029,7 @@ async def test_passthrough_body_cannot_forge_budget_reservation(): user_api_key_dict, parsed_body={"litellm_metadata": {"user_api_key_budget_reservation": forged}}, ) - assert ( - kwargs["litellm_params"]["metadata"]["user_api_key_budget_reservation"] is None - ) + assert kwargs["litellm_params"]["metadata"]["user_api_key_budget_reservation"] is None increment_spend_counters = await _track_cost_for_passthrough_kwargs(kwargs) @@ -5391,9 +5037,7 @@ async def test_passthrough_body_cannot_forge_budget_reservation(): assert increment_spend_counters.await_args.kwargs["budget_reservation"] is None -async def _drive_streaming_pass_through( - upstream_content_type, chunk_delay_seconds, client_asked_for_stream=True -): +async def _drive_streaming_pass_through(upstream_content_type, chunk_delay_seconds, client_asked_for_stream=True): """Drive pass_through_request against an upstream that stalls before its first byte. ``client_asked_for_stream`` picks which of pass_through_request's two streaming @@ -5405,22 +5049,14 @@ async def _drive_streaming_pass_through( ) with ExitStack() as stack: - mock_proxy_logging = stack.enter_context( - patch("litellm.proxy.proxy_server.proxy_logging_obj") - ) + mock_proxy_logging = stack.enter_context(patch("litellm.proxy.proxy_server.proxy_logging_obj")) mock_get_client = stack.enter_context( - patch( - "litellm.proxy.pass_through_endpoints.pass_through_endpoints.get_async_httpx_client" - ) - ) - mock_chunk_processor = stack.enter_context( - patch.object(PassThroughStreamingHandler, "chunk_processor") + patch("litellm.proxy.pass_through_endpoints.pass_through_endpoints.get_async_httpx_client") ) + mock_chunk_processor = stack.enter_context(patch.object(PassThroughStreamingHandler, "chunk_processor")) mock_proxy_logging.pre_call_hook = AsyncMock( - return_value={"model": "claude-3", "stream": True} - if client_asked_for_stream - else {"model": "claude-3"} + return_value={"model": "claude-3", "stream": True} if client_asked_for_stream else {"model": "claude-3"} ) mock_proxy_logging.post_call_failure_hook = AsyncMock() mock_proxy_logging.post_call_response_headers_hook = AsyncMock(return_value={}) @@ -5510,9 +5146,7 @@ async def test_pass_through_binary_event_stream_is_never_given_an_sse_comment(): @pytest.mark.asyncio @pytest.mark.parametrize("configured_interval, expect_ping", [(0.05, True), (None, False)]) -async def test_pass_through_route_pings_while_the_upstream_call_is_still_running( - configured_interval, expect_ping -): +async def test_pass_through_route_pings_while_the_upstream_call_is_still_running(configured_interval, expect_ping): """The upstream withholds its response headers until its first token, so the whole time-to-first-token is spent inside pass_through_request with nothing on the wire (issue #34819).""" @@ -5542,9 +5176,7 @@ async def test_pass_through_route_pings_while_the_upstream_call_is_still_running ) ) stack.enter_context(patch(f"{module}.pass_through_request", slow_pass_through)) - stack.enter_context( - patch.object(litellm, "sse_keepalive_ping_interval_seconds", configured_interval) - ) + stack.enter_context(patch.object(litellm, "sse_keepalive_ping_interval_seconds", configured_interval)) endpoint_func = create_pass_through_route( endpoint="/v1/messages", @@ -5571,3 +5203,147 @@ async def test_pass_through_route_pings_while_the_upstream_call_is_still_running assert (collected[0] == b": ping\n\n") is expect_ping assert collected[-1] in (MESSAGE_START_SSE_FRAME, MESSAGE_START_SSE_FRAME.decode()) + + +def test_passthrough_carries_the_per_model_budgets(): + """ + Native passthrough builds its logging metadata from + StandardLoggingUserAPIKeyMetadata, which has no budget field, and never calls + add_litellm_data_to_request. Without these three keys the post-call increment + exits early, so a /bedrock/... request is costed but its per-model counter is + never written: the budget reports zero forever and enforces nothing. + """ + key_budget = {"claude-opus-4-8": {"budget_limit": 1.0, "time_period": "18h"}} + user_budget = {"claude-opus-4-8": {"budget_limit": 2.0, "time_period": "1mo"}} + end_user_budget = {"claude-opus-4-8": {"budget_limit": 3.0, "time_period": "1d"}} + + kwargs = _passthrough_kwargs_for_reservation( + UserAPIKeyAuth( + token="hash", + user_id="u-1", + model_max_budget=key_budget, + user_model_max_budget=user_budget, + end_user_model_max_budget=end_user_budget, + ) + ) + + metadata = kwargs["litellm_params"]["metadata"] + assert metadata["user_api_key_model_max_budget"] == key_budget + assert metadata["user_api_key_user_model_max_budget"] == user_budget + assert metadata["user_api_key_end_user_model_max_budget"] == end_user_budget + + +def test_passthrough_budget_metadata_cannot_be_forged_by_the_request_body(): + """ + These keys decide budget enforcement, so a caller-supplied body must not be + able to raise its own cap. They are set after the client metadata merge for + the same reason user_api_key and the parent span are. + """ + key_budget = {"claude-opus-4-8": {"budget_limit": 1.0, "time_period": "18h"}} + + kwargs = _passthrough_kwargs_for_reservation( + UserAPIKeyAuth(token="hash", user_id="u-1", model_max_budget=key_budget), + parsed_body={ + "litellm_metadata": { + "user_api_key_model_max_budget": {"claude-opus-4-8": {"budget_limit": 999999.0, "time_period": "18h"}} + } + }, + ) + + metadata = kwargs["litellm_params"]["metadata"] + assert metadata["user_api_key_model_max_budget"] == key_budget + + +def _marked_pass_through_endpoint(): + """An endpoint carrying the marker ``create_pass_through_route`` sets.""" + from litellm.types.passthrough_endpoints.pass_through_endpoints import ( + LITELLM_PASS_THROUGH_ENDPOINT_MARKER, + ) + + def _endpoint(): # pragma: no cover - identity only + return None + + setattr(_endpoint, LITELLM_PASS_THROUGH_ENDPOINT_MARKER, True) # noqa: B010 # name is a module constant + return _endpoint + + +def test_user_defined_passthrough_is_neither_tracked_nor_enforced(): + """ + `get_model_from_request` returns None for a user-defined pass-through on + purpose: the body is forwarded verbatim, so its `model` names an UPSTREAM + model rather than a LiteLLM-managed one, and enforcing key/team allowlists + against it would reject valid requests. Enforcement is therefore skipped + on those routes. + + Attaching the budget metadata anyway would charge a counter that nothing on + that route can refuse, and would attribute the spend to a budget the operator + scoped to a LiteLLM model that merely shares the name. Tracking and + enforcement have to agree: both on for the built-in provider routes, both off + here. + """ + kwargs = _passthrough_kwargs_for_reservation( + UserAPIKeyAuth( + token="hash", + user_id="u-1", + model_max_budget={"claude-opus-4-8": {"budget_limit": 1.0, "time_period": "18h"}}, + ), + user_defined_route=True, + ) + + metadata = kwargs["litellm_params"]["metadata"] + for field in ( + "user_api_key_model_max_budget", + "user_api_key_user_model_max_budget", + "user_api_key_end_user_model_max_budget", + ): + assert field not in metadata, f"{field} was attached on a route that never enforces it" + + +@pytest.mark.parametrize( + "handler_name", + [ + "anthropic_proxy_route", + "bedrock_proxy_route", + "gemini_proxy_route", + "cohere_proxy_route", + "vllm_proxy_route", + "mistral_proxy_route", + ], +) +def test_builtin_provider_routes_do_not_carry_the_user_defined_marker(handler_name): + """ + The budget metadata is attached only when the dispatched endpoint is NOT a + user-defined pass-through, so the built-in provider handlers must not carry + that marker or native provider spend would stop being tracked and enforced. + + These handlers DO call `create_pass_through_route` internally, and that + factory sets the marker on what it returns. But the result is awaited + immediately rather than registered, so FastAPI puts the decorated handler in + `request.scope["endpoint"]`, and that is what the marker check reads. This + test pins the distinction between calling the factory and being dispatched as + its product, which is easy to misread from a grep alone. + """ + from litellm.proxy.pass_through_endpoints import llm_passthrough_endpoints + from litellm.types.passthrough_endpoints.pass_through_endpoints import ( + LITELLM_PASS_THROUGH_ENDPOINT_MARKER, + ) + + handler = getattr(llm_passthrough_endpoints, handler_name) + assert getattr(handler, LITELLM_PASS_THROUGH_ENDPOINT_MARKER, False) is False, ( + f"{handler_name} is marked as a user-defined pass-through, so per-model budget " + "metadata would be skipped and native provider spend would go untracked" + ) + + +def test_the_marker_check_distinguishes_the_two_route_kinds(): + """Positive control: the factory's product IS marked, so the check can discriminate.""" + from litellm.proxy.auth.auth_utils import request_dispatched_to_pass_through_endpoint + from litellm.proxy.pass_through_endpoints import llm_passthrough_endpoints + + marked = MagicMock(spec=Request) + marked.scope = {"endpoint": _marked_pass_through_endpoint()} + assert request_dispatched_to_pass_through_endpoint(marked) is True + + builtin = MagicMock(spec=Request) + builtin.scope = {"endpoint": llm_passthrough_endpoints.anthropic_proxy_route} + assert request_dispatched_to_pass_through_endpoint(builtin) is False diff --git a/ui/litellm-dashboard/src/app/(dashboard)/users/_components/BulkEditUsers.tsx b/ui/litellm-dashboard/src/app/(dashboard)/users/_components/BulkEditUsers.tsx index b3326df1ff8..459e3fd8c92 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/users/_components/BulkEditUsers.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/users/_components/BulkEditUsers.tsx @@ -10,6 +10,7 @@ import { Checkbox } from "@/components/ui/checkbox"; import { Dialog, DialogContent, DialogHeader, DialogTitle } from "@/components/ui/dialog"; import { Separator } from "@/components/ui/separator"; import { Table, TableBody, TableCell, TableHead, TableHeader, TableRow } from "@/components/ui/table"; +import useAuthorized from "@/app/(dashboard)/hooks/useAuthorized"; interface BulkEditUserModalProps { open: boolean; @@ -36,6 +37,7 @@ const BulkEditUserModal: React.FC = ({ userModels, allowAllUsers = false, }) => { + const { premiumUser } = useAuthorized(); const [loading, setLoading] = useState(false); const [selectedTeams, setSelectedTeams] = useState([]); const [teamBudget, setTeamBudget] = useState(null); @@ -362,6 +364,7 @@ const BulkEditUserModal: React.FC = ({ userModels={userModels} possibleUIRoles={possibleUIRoles} isBulkEdit={true} + premiumUser={premiumUser === true} /> {loading && ( diff --git a/ui/litellm-dashboard/src/app/(dashboard)/users/_components/user_edit_view.test.tsx b/ui/litellm-dashboard/src/app/(dashboard)/users/_components/user_edit_view.test.tsx index 0ca68cc665e..8fb94ce477e 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/users/_components/user_edit_view.test.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/users/_components/user_edit_view.test.tsx @@ -1,4 +1,4 @@ -import { cleanup, screen, waitFor } from "@testing-library/react"; +import { cleanup, fireEvent, screen, waitFor } 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"; @@ -612,6 +612,125 @@ describe("UserEditView", () => { expect(onSubmit).not.toHaveBeenCalled(); }); + // /user/new validates model_max_budget behind an enterprise license, so a + // form that re-sends what is already stored turns an unrelated edit into a + // 400 on a proxy without one. + describe("per-model budgets", () => { + const withStoredBudgets = { + ...MOCK_USER_DATA, + user_info: { + ...MOCK_USER_DATA.user_info, + model_max_budget: { "gpt-4": { budget_limit: 5, time_period: "30d" } }, + }, + }; + + it("should leave model_max_budget out of an edit that did not touch it", async () => { + const payload = await submittedPayload({ userData: withStoredBudgets, premiumUser: true }); + + expect(payload).not.toHaveProperty("model_max_budget"); + }); + + // The proxy stores model_max_budget as a plain dict, exactly as the client + // sent it, and BudgetConfig documents the max_budget/budget_duration + // spelling. A row hydrated from the spelling the editor does not read mounts + // with an empty cap, and every edit re-emits ALL rows, so touching one + // model's budget silently deletes another's. + it("should keep a row stored under the BudgetConfig aliases when a sibling row is edited", async () => { + const onSubmit = vi.fn(); + renderWithProviders( + , + ); + + const [aliasRow, canonicalRow] = await screen.findAllByPlaceholderText("Max spend ($)"); + expect(aliasRow).toHaveValue(5); + + fireEvent.change(canonicalRow, { target: { value: "3" } }); + await userEvent.click(screen.getByRole("button", { name: /save changes/i })); + + await waitFor(() => { + expect(onSubmit).toHaveBeenCalled(); + }); + expect(onSubmit.mock.calls[0][0].model_max_budget).toEqual({ + "gpt-4": { budget_limit: 5, time_period: "30d" }, + "gpt-3.5-turbo": { budget_limit: 3, time_period: "1h" }, + }); + }); + + // The effect already re-seeds the form on a userData change, so that change + // does happen while this component stays mounted. The editor holds its rows + // in state seeded once, so without a matching re-seed the rows on screen + // keep describing the previously loaded user and a save overwrites theirs. + it("re-seeds the editor when a different user is loaded", async () => { + const withBudget = (limit: number, id: string) => ({ + ...MOCK_USER_DATA, + user_id: id, + user_info: { + ...MOCK_USER_DATA.user_info, + model_max_budget: { "gpt-4": { budget_limit: limit, time_period: "1h" } }, + }, + }); + + const { rerender } = renderWithProviders( + , + ); + expect(await screen.findByPlaceholderText("Max spend ($)")).toHaveValue(5); + + rerender(); + + expect(await screen.findByPlaceholderText("Max spend ($)")).toHaveValue(99); + }); + + // BulkEditUsers copies a fixed field list into its payload and never reads + // model_max_budget, so an editor rendered here would take input and throw + // it away. It also has no single stored budget to diff against, since its + // userData stands in for every selected user. + it("does not offer the editor in bulk edit, where the value would be discarded", async () => { + renderWithProviders( + , + ); + + await screen.findByRole("button", { name: /save changes/i }); + expect(screen.queryByPlaceholderText("Max spend ($)")).not.toBeInTheDocument(); + }); + + it("should lock the editor when the proxy has no enterprise license", async () => { + renderWithProviders(); + + expect(await screen.findByPlaceholderText("Max spend ($)")).toBeDisabled(); + }); + + it("should leave the editor usable when the proxy has one", async () => { + renderWithProviders(); + + expect(await screen.findByPlaceholderText("Max spend ($)")).toBeEnabled(); + }); + }); + it("should send an empty-string metadata through untouched rather than as an object", async () => { const onSubmit = vi.fn(); renderWithProviders(); diff --git a/ui/litellm-dashboard/src/app/(dashboard)/users/_components/user_edit_view.tsx b/ui/litellm-dashboard/src/app/(dashboard)/users/_components/user_edit_view.tsx index 5612f5cd5f6..ed0c08adf38 100644 --- a/ui/litellm-dashboard/src/app/(dashboard)/users/_components/user_edit_view.tsx +++ b/ui/litellm-dashboard/src/app/(dashboard)/users/_components/user_edit_view.tsx @@ -2,6 +2,9 @@ import React, { useMemo, useState } from "react"; import { z } from "zod/v4"; import { all_admin_roles } from "@/utils/roles"; import BudgetDurationDropdown from "@/components/common_components/budget_duration_dropdown"; +import { ModelMaxBudget, ModelMaxBudgetField } from "@/components/key_team_helpers/ModelMaxBudgetEditor"; +import { modelMaxBudgetUpdate } from "@/components/key_team_helpers/modelMaxBudgetPayload"; +import { useSeededState } from "@/components/key_team_helpers/useSeededState"; import { getModelDisplayName } from "@/components/key_team_helpers/fetch_available_models_team_key"; import MCPServerSelector from "@/components/mcp_server_management/MCPServerSelector"; import MCPToolPermissions from "@/components/mcp_server_management/MCPToolPermissions"; @@ -30,6 +33,7 @@ interface UserEditViewProps { possibleUIRoles: Record> | null; isBulkEdit?: boolean; objectPermission?: ObjectPermission | null; + premiumUser?: boolean; } const MCP_SELECTION_SHAPE = z.object({ @@ -135,9 +139,14 @@ export function UserEditView({ possibleUIRoles, isBulkEdit = false, objectPermission, + premiumUser = false, }: UserEditViewProps) { const canEditMcpPermissions = !isBulkEdit && all_admin_roles.includes(userRole || ""); const [unlimitedBudget, setUnlimitedBudget] = useState(false); + const [modelMaxBudget, setModelMaxBudget] = useSeededState( + userData.user_id, + () => userData.user_info?.model_max_budget ?? {}, + ); const schema = useMemo(() => budgetSchema(unlimitedBudget), [unlimitedBudget]); const form = useZodForm(schema, { defaultValues: toFormValues(userData, objectPermission, isBulkEdit, canEditMcpPermissions), @@ -162,9 +171,11 @@ export function UserEditView({ return; } + const modelBudgets = modelMaxBudgetUpdate(modelMaxBudget, userData.user_info?.model_max_budget); onSubmit({ ...values, ...("metadata" in values ? { metadata: metadata.value } : {}), + ...(modelBudgets !== undefined && { model_max_budget: modelBudgets }), max_budget: unlimitedBudget || values.max_budget === "" || values.max_budget === undefined ? null : values.max_budget, }); @@ -282,6 +293,20 @@ export function UserEditView({ {({ id, value, onChange }) => } + {/* Bulk edit forwards a fixed field list and has no single stored budget to + diff against, so the editor would silently discard whatever was typed. */} + {!isBulkEdit && ( + + )} + {({ ref, value, ...control }) => (