Skip to content

Fix: tool call streaming when both reasoning and tool parsers are enabled #28297 #28330

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 2 commits into
base: main
Choose a base branch
from
+12 −2
Open

Fix: tool call streaming when both reasoning and tool parsers are enabled #28297 #28330

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
2df2ff4
Fix: tool call streaming when both reasoning and tool parsers are ena…
baonudesifeizhai Nov 7, 2025
fe3814f
change for reviewer advice
baonudesifeizhai Nov 8, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 12 additions & 2 deletions vllm/entrypoints/openai/serving_chat.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -946,7 +946,15 @@ async def chat_completion_stream_generator(
assert added_content_delta_arr is not None
assert reasoning_end_arr is not None
output_token_ids = as_list(output.token_ids)
if not reasoning_end_arr[i]:

# Process reasoning only if tool call tokens are not present
# This handles cases where models output tool calls without
# reasoning content (e.g., Qwen3-VL with hermes tool parser)
if not reasoning_end_arr[i] and (
not tool_parser
or not hasattr(tool_parser, "tool_call_start_token")
or tool_parser.tool_call_start_token not in current_text
):
delta_message = (
reasoning_parser.extract_reasoning_content_streaming(
previous_text,
Expand Down Expand Up @@ -992,8 +1000,10 @@ async def chat_completion_stream_generator(
else:
current_text = ""

# handle tool calls only after reasoning is done,
# handle tool calls after reasoning is done or when tool call
# tokens are detected
else:
reasoning_end_arr[i] = True
delta_token_ids = output_token_ids
# First time to tool call,
# add the remaining text and token ids
Expand Down