A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
[gpt-oss][bugfix] remove logic to require resp_ in ResponseAPI (#25428)
Signed-off-by: Andrew Xia <axia@meta.com>
Andrew Xia committed
Sep 23, 2025 at 15:46 UTC
95bc60e4cb42029dbe7f1b7c568d5147f0ea1f2d
2 files changed
+1
-15
tests/entrypoints/openai/test_response_api_with_harmony.py
+1
@@ -523,6 +523,7 @@ async def test_function_calling(client: OpenAI, model_name: str):
523
input="What's the weather like in Paris today?",
524
tools=tools,
525
temperature=0.0,
526
+ extra_body={"request_id": "test_function_calling_non_resp"},
527
)
528
assert response is not None
529
assert response.status == "completed"
vllm/entrypoints/openai/serving_responses.py
-15
@@ -235,8 +235,6 @@ class OpenAIServingResponses(OpenAIServing):
235
# Handle the previous response ID.
236
prev_response_id = request.previous_response_id
237
if prev_response_id is not None:
238
- if not prev_response_id.startswith("resp_"):
239
- return self._make_invalid_id_error(prev_response_id)
238
async with self.response_store_lock:
239
prev_response = self.response_store.get(prev_response_id)
240
if prev_response is None:
@@ -924,9 +922,6 @@ class OpenAIServingResponses(OpenAIServing):
922
stream: Optional[bool],
923
) -> Union[ErrorResponse, ResponsesResponse, AsyncGenerator[
924
StreamingResponsesResponse, None]]:
927
- if not response_id.startswith("resp_"):
928
- return self._make_invalid_id_error(response_id)
929
-
925
async with self.response_store_lock:
926
response = self.response_store.get(response_id)
927
@@ -944,9 +939,6 @@ class OpenAIServingResponses(OpenAIServing):
939
self,
940
response_id: str,
941
) -> Union[ErrorResponse, ResponsesResponse]:
947
- if not response_id.startswith("resp_"):
948
- return self._make_invalid_id_error(response_id)
949
-
942
async with self.response_store_lock:
943
response = self.response_store.get(response_id)
944
if response is None:
@@ -972,13 +964,6 @@ class OpenAIServingResponses(OpenAIServing):
964
response_id)
965
return response
966
975
- def _make_invalid_id_error(self, response_id: str) -> ErrorResponse:
976
- return self.create_error_response(
977
- err_type="invalid_request_error",
978
- message=(f"Invalid 'response_id': '{response_id}'. "
979
- "Expected an ID that begins with 'resp'."),
980
- )
981
-
967
def _make_not_found_error(self, response_id: str) -> ErrorResponse:
968
return self.create_error_response(
969
err_type="invalid_request_error",