[Bugfix][Parser] Pass token IDs to parser.parse() in Responses API and batch serving (#46843)
Signed-off-by: Ben Browning <bbrownin@redhat.com>
Ben Browning committed
Jun 26, 2026 at 15:29 UTC
dccb412e2c72a0c147166c14b25c01f045a74163
3 files changed
+3
vllm/entrypoints/openai/chat_completion/batch_serving.py
+1
@@ -267,6 +267,7 @@ class OpenAIServingChatBatch(OpenAIServingChat):
267
reasoning, content, _ = parser.parse(
268
output.text,
269
request=request, # type: ignore[arg-type]
270
+ model_output_token_ids=output.token_ids,
271
)
272
if not request.include_reasoning:
273
reasoning = None
vllm/entrypoints/openai/responses/context.py
+1
@@ -346,6 +346,7 @@ class ParsableContext(ConversationContext):
346
completion.text,
347
self.request,
348
enable_auto_tools=self.enable_auto_tools,
349
+ model_output_token_ids=completion.token_ids,
350
)
351
self.response_messages.extend(
352
build_response_output_items(
vllm/entrypoints/openai/responses/serving.py
+1
@@ -1063,6 +1063,7 @@ class OpenAIServingResponses(OpenAIServing):
1063
final_output.text,
1064
request,
1065
enable_auto_tools=self.enable_auto_tools,
1066
+ model_output_token_ids=final_output.token_ids,
1067
)
1068
return build_response_output_items(
1069
reasoning=reasoning,