diff --git a/litellm/llms/custom_httpx/llm_http_handler.py b/litellm/llms/custom_httpx/llm_http_handler.py index 973d0336ee9..7b808cf2be2 100644 --- a/litellm/llms/custom_httpx/llm_http_handler.py +++ b/litellm/llms/custom_httpx/llm_http_handler.py @@ -2926,14 +2926,18 @@ class BaseLLMHTTPHandler: else: sync_httpx_client = client - headers = _merge_headers_case_insensitive( # rebind-ok: validation replaces the method's incoming headers - responses_api_provider_config.validate_environment( - headers=_client_request_headers(response_api_optional_request_params, extra_headers), + headers = { # rebind-ok: validation replaces the method's incoming headers + **responses_api_provider_config.validate_environment( + headers=( + _client_request_headers(response_api_optional_request_params, extra_headers) + if custom_llm_provider == "chatgpt" + else response_api_optional_request_params.get("extra_headers", {}) or {} + ), model=model, litellm_params=litellm_params, ), - extra_headers if custom_llm_provider != "chatgpt" else None, - ) + **((extra_headers or {}) if custom_llm_provider != "chatgpt" else {}), + } # Check if streaming is requested stream = response_api_optional_request_params.get("stream", False) @@ -3121,15 +3125,19 @@ class BaseLLMHTTPHandler: else: async_httpx_client = client - validated_headers: Final = _merge_headers_case_insensitive( - await asyncio.to_thread( + validated_headers: Final = { + **await asyncio.to_thread( responses_api_provider_config.validate_environment, - headers=_client_request_headers(response_api_optional_request_params, extra_headers), + headers=( + _client_request_headers(response_api_optional_request_params, extra_headers) + if custom_llm_provider == "chatgpt" + else response_api_optional_request_params.get("extra_headers", {}) or {} + ), model=model, litellm_params=litellm_params, ), - extra_headers if custom_llm_provider != "chatgpt" else None, - ) + **((extra_headers or {}) if custom_llm_provider != "chatgpt" else {}), + } # Check if streaming is requested stream = response_api_optional_request_params.get("stream", False) diff --git a/litellm/router.py b/litellm/router.py index 98315747d52..acabb5f41c3 100644 --- a/litellm/router.py +++ b/litellm/router.py @@ -4813,11 +4813,17 @@ class Router: ) declared_provider: Final = deployment_context.get("custom_llm_provider") declared_model: Final = deployment_context.get("model") - forwarded_authorization: Final = ProviderSpecificHeaderUtils.forwarded_authorization( - request_context.get("provider_specific_header"), + deployment_provider: Final = ( declared_provider if isinstance(declared_provider, str) - else inferred_provider(declared_model if isinstance(declared_model, str) else None), + else inferred_provider(declared_model if isinstance(declared_model, str) else None) + ) + forwarded_authorization: Final = ( + ProviderSpecificHeaderUtils.forwarded_authorization( + request_context.get("provider_specific_header"), deployment_provider + ) + if deployment_provider == "chatgpt" + else None ) if is_clientside_credential(request_kwargs=kwargs) or forwarded_authorization is not None: deployment_pydantic_obj: Final = self._handle_clientside_credential( diff --git a/tests/unit/llms/custom_httpx/test_llm_http_handler.py b/tests/unit/llms/custom_httpx/test_llm_http_handler.py index bbcd7f70e63..f7850aaf9a1 100644 --- a/tests/unit/llms/custom_httpx/test_llm_http_handler.py +++ b/tests/unit/llms/custom_httpx/test_llm_http_handler.py @@ -5654,3 +5654,44 @@ def test_vector_store_retrieve_handler_sends_id_before_azure_api_version_query() sent: Final = received[0] assert str(sent.url) == "https://x.openai.azure.com/openai/vector_stores/vs_abc?api-version=2025-04-01-preview" assert len(received) == 1 + + +@pytest.mark.asyncio +@pytest.mark.parametrize("is_async", (False, True)) +@pytest.mark.parametrize("provider", ("openai", "azure")) +async def test_non_chatgpt_responses_preserves_caller_headers_after_validation( + is_async: bool, provider: str, respx_mock: respx.MockRouter, +) -> None: + route: Final = respx_mock.post(url__regex=r"https://responses\.example\.com/.*").respond( + 200, + json={ + "id": "resp_headers", "object": "response", "created_at": 1, + "model": "gpt-6-luna", "status": "completed", "output": [], + "parallel_tool_calls": False, "tool_choice": "auto", "tools": [], + }, + ) + client: Final = AsyncHTTPHandler(transport=httpx.AsyncHTTPTransport()) if is_async else HTTPHandler(client=httpx.Client()) + kwargs: Final = { + "model": f"{provider}/gpt-6-luna", + "input": "hello", + "api_key": "fake-configured-key", + "api_base": "https://responses.example.com/v1", + "api_version": "2025-04-01-preview", + "extra_headers": {"Authorization": "Bearer fake-caller-key", "content-type": "application/vnd.example+json"}, + "client": client, + } + try: + response: Final = await litellm.aresponses(**kwargs) if is_async else litellm.responses(**kwargs) + assert response.status == "completed" + assert route.calls[0].request.headers["authorization"] == "Bearer fake-caller-key" + expected_content_types: Final = ( + ["application/json", "application/vnd.example+json"] + if provider == "openai" + else ["application/vnd.example+json"] + ) + assert route.calls[0].request.headers.get_list("content-type") == expected_content_types + finally: + if isinstance(client, AsyncHTTPHandler): + await client.close() + else: + client.client.close() diff --git a/tests/unit/test_router/test_router.py b/tests/unit/test_router/test_router.py index 70376fc4069..28fffb29a1a 100644 --- a/tests/unit/test_router/test_router.py +++ b/tests/unit/test_router/test_router.py @@ -27,6 +27,7 @@ from pydantic import BaseModel, TypeAdapter import litellm from litellm import APIConnectionError, Router +from litellm.llms.custom_httpx.http_handler import AsyncHTTPHandler from litellm._logging import verbose_logger, verbose_router_logger from litellm.caching import RedisCache, RedisClusterCache from litellm.caching.caching import DualCache @@ -30158,3 +30159,48 @@ def test_request_carries_image_sees_chat_parts_and_responses_input_items(): assert litellm.Router._request_carries_image(None, {"input": _IMAGE_RESPONSES_INPUT}) is True assert litellm.Router._request_carries_image([{"role": "user", "content": "Say hi"}], {}) is False assert litellm.Router._request_carries_image(None, {"input": "Say hi"}) is False + + +@pytest.mark.asyncio +async def test_anthropic_forwarded_oauth_preserves_configured_deployment( + respx_mock: respx.MockRouter, +) -> None: + route: Final = respx_mock.post("https://api.anthropic.com/v1/messages").respond( + 200, + json={ + "id": "msg-oauth", + "type": "message", + "role": "assistant", + "model": "claude-sonnet-5-5", + "content": [{"type": "text", "text": "subscription reply"}], + "stop_reason": "end_turn", + "stop_sequence": None, + "usage": {"input_tokens": 2, "output_tokens": 2}, + }, + ) + router: Final = Router( + model_list=[{ + "model_name": "subscription", + "litellm_params": {"model": "anthropic/claude-sonnet-5-5", "api_key": "fake-saved-key"}, + "model_info": {"id": "configured-anthropic-deployment"}, + }], + num_retries=0, + ) + client: Final = AsyncHTTPHandler(transport=httpx.AsyncHTTPTransport()) + try: + for token in ("sk-ant-oat-fake-client-a", "sk-ant-oat-fake-client-b"): + response: Final = await router.acompletion( + model="subscription", + messages=[{"role": "user", "content": "hello"}], + client=client, + provider_specific_header={ + "custom_llm_provider": "anthropic", + "extra_headers": {"authorization": f"Bearer {token}"}, + }, + ) + assert response.choices[0].message.content == "subscription reply" + assert response._hidden_params["model_id"] == "configured-anthropic-deployment" + assert route.calls[-1].request.headers["authorization"] == f"Bearer {token}" + assert "x-api-key" not in route.calls[-1].request.headers + finally: + await client.close()