#24007generating chat title will block `/v1/chat-messages` streaming response

Self Checks

• I have read the Contributing Guide and Language Policy.
• This is only for bug report, if you would like to ask a question, please head to Discussions.
• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report, otherwise it will be closed.
• 【中文用户 & Non English User】请使用英语提交，否则会被关闭 ：）
• Please do not modify this template :) and fill in all the required fields.

Dify version

1.7.1

Cloud or Self Hosted

Self Hosted (Docker)

Steps to reproduce

It's related to my last issue #23311 At first I found the problem was with the model provider plugin I used. When I used huawei cloud maas provider, the issue appeared, but using openai-compatible is fine. And I found it from plugin daemon logs that for each new conversation, the llm/invoke is triggered twice. One is normal, another takes much longer time which is almost the same showing on Postman.

Thanks to this issue https://github.com/langgenius/dify-plugin-daemon/issues/329, I set up a proxy on plugin daemon and I finally got what the second invoke is used for.

It used deepseek-r1 to generate chat title which took around 25 seconds to finish. So I can resolve this by using a non-cot model as system llm model, but in general the new conversation should not hang once the workflow is done.

✔️ Expected Behavior

The chat title request should not block the /v1/chat-messages on streaming mode. Same test for /api/chat-messages, no similar issue found.

❌ Actual Behavior

The /v1/chat-messages streaming cannot close until the chat title request is finished.