lm-sys / lm-sys/FastChat

Qwen-3 produces inconsistent output on the Web UI (works on api)

Open
#3,724 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
39.5k
Forks
4.8k
PR merge metrics
No merged PRs in 30d

Description

It seems that the reasoning steps get killed on the web ui. If I use FastChat's openAI API, I can see that the model is working correctly, and I can see the reasoning happening.

Here it is using the fast chat gradio_web_server :
Image
And this is the output when using open-webui accessing the same FastChat instance:
Image

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Reproduce the difference between FastChat's gradio_web_server and the same FastChat instance accessed through open-webui, using Qwen-3. Start by comparing how reasoning output is returned by the FastChat OpenAI API and displayed in the web UI. Done means the web UI preserves the reasoning output consistently with the API.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
api, frontend
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.