fix(responses): forward allowed_openai_params through the chat completions bridge (#35885)

Resolves #35878

Co-authored-by: devin-ai-integration[bot] <158243242+devin-ai-integration[bot]@users.noreply.github.com>

litellm/responses/main.py

@@ -1064,6 +1064,7 @@ def responses(
                 extra_headers=extra_headers,
                 extra_body=extra_body,
                 timeout=timeout if timeout is not None else request_timeout,
+                allowed_openai_params=allowed_openai_params,
                 **kwargs,
             )
 

tests/test_litellm/responses/test_responses_api_bridge_flag.py

@@ -16,6 +16,7 @@ sys.path.insert(
 
 import litellm
 from litellm.types.llms.openai import ResponseAPIUsage, ResponsesAPIResponse
+from litellm.types.utils import Choices, Message, ModelResponse, Usage
 
 
 class TestUseResponsesApiBridgeFlag:
@@ -130,6 +131,44 @@ class TestUseResponsesApiBridgeFlag:
 
         mock_bridge_handler.assert_called_once()
 
+    @patch("litellm.acompletion")
+    @patch(
+        "litellm.responses.main.ProviderConfigManager.get_provider_responses_api_config"
+    )
+    async def test_allowed_openai_params_forwarded_through_bridge(
+        self, mock_get_config, mock_acompletion
+    ):
+        """allowed_openai_params is a named param of responses(), so it must be
+        explicitly forwarded to the bridge; otherwise litellm.acompletion raises
+        UnsupportedParamsError for params the caller explicitly allowed."""
+        mock_get_config.return_value = litellm.OpenAIResponsesAPIConfig()
+        mock_acompletion.return_value = ModelResponse(
+            id="chatcmpl_123",
+            model="openai/my-custom-model",
+            choices=[
+                Choices(
+                    index=0,
+                    message=Message(role="assistant", content="Answer"),
+                    finish_reason="stop",
+                )
+            ],
+            usage=Usage(prompt_tokens=10, completion_tokens=5, total_tokens=15),
+        )
+
+        await litellm.aresponses(
+            model="openai/my-custom-model",
+            input="Hello",
+            use_chat_completions_api=True,
+            allowed_openai_params=["reasoning_effort"],
+            reasoning={"effort": "high"},
+            litellm_logging_obj=MagicMock(),
+        )
+
+        mock_acompletion.assert_called_once()
+        assert mock_acompletion.call_args.kwargs.get("allowed_openai_params") == [
+            "reasoning_effort"
+        ]
+
     @patch("litellm.responses.file_search.emulated_handler._call_aresponses")
     @patch(
         "litellm.responses.main.ProviderConfigManager.get_provider_responses_api_config"