github-starred/open-webui
2
0
Fork 0
mirror of https://github.com/open-webui/open-webui.git synced 2026-08-30 17:25:30 -05:00
Code Issues 2.4k Packages Projects Releases 129 Wiki Activity

issue: Non-closing think section with latest openAI API backend with reasoning_parser #5618

New Issue
Closed
opened 2025-11-11 16:26:27 -06:00 by GiteaMirror · 2 comments
GiteaMirror commented 2025-11-11 16:26:27 -06:00
Owner
Copy Link
Copy Source

Originally created by @jingyibo123 on GitHub (Jun 23, 2025).

Check Existing Issues

  • I have searched the existing issues and discussions.
  • I am using the latest version of Open WebUI.

Installation Method

Git Clone

Open WebUI Version

0.6.15

Ollama Version (if applicable)

No response

Operating System

Ubuntu 20.04

Browser (if applicable)

No response

Confirmation

  • I have read and followed all instructions in README.md.
  • I am using the latest version of both Open WebUI and Ollama.
  • I have included the browser console logs.
  • I have included the Docker container logs.
  • I have provided every relevant configuration, setting, and environment variable used in my setup.
  • I have clearly listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc).
  • I have documented step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation. My steps:
  • Start with the initial platform/version/OS and dependencies used,
  • Specify exact install/launch/configure commands,
  • List URLs visited, user input (incl. example values/emails/passwords if needed),
  • Describe all options and toggles enabled or changed,
  • Include any files or environmental changes,
  • Identify the expected and actual result at each stage,
  • Ensure any reasonably skilled user can follow and hit the same issue.

Expected Behavior

Expected behavior: show reasoning_content in the frontend folded think-section, and when reasoning_content is over and content starts, finish the think-section and show content outside..

Actual Behavior

When working on think-enabled models such as qwen3 on latest openAI server deployed with lmdeploy, which returns first reasoning_content then content without </think> clouse, the open-webui frontend shows a ever-lasting thinking section including both the reasoning_content and the actual content.

lmdeploy's API's behavior ( stream response):

... content=<think>...
... reasoning_content="XX"...
... reasoning_content="XX"...
... reasoning_content="XX"...
... content="XX"...
... content="XX"...

vllm's API's behavior ( stream response):

... content=""...
... reasoning_content="XX"...
... reasoning_content="XX"...
... reasoning_content="XX"...
... content="XX"...
... content="XX"...

without reasoning enabled:

... content="<think>"...
... content="XX"...
... content="XX"...
... content="</think>"...
... content="XX"...
... content="XX"...

Open-webuiworks OK when not enabling reasoning or enabled with vllm, NOT with deploy

Steps to Reproduce

Deploy mode with lmdeploy

lmdeploy serve api_server \
    Qwen/Qwen3-32B \
    --reasoning-parser qwen-qwq \
    --tool-call-parser qwen

Logs & Screenshots

No error in logs.

Additional Information

No response

Originally created by @jingyibo123 on GitHub (Jun 23, 2025). ### Check Existing Issues - [x] I have searched the existing issues and discussions. - [x] I am using the latest version of Open WebUI. ### Installation Method Git Clone ### Open WebUI Version 0.6.15 ### Ollama Version (if applicable) _No response_ ### Operating System Ubuntu 20.04 ### Browser (if applicable) _No response_ ### Confirmation - [x] I have read and followed all instructions in `README.md`. - [x] I am using the latest version of **both** Open WebUI and Ollama. - [x] I have included the browser console logs. - [x] I have included the Docker container logs. - [x] I have **provided every relevant configuration, setting, and environment variable used in my setup.** - [x] I have clearly **listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup** (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc). - [x] I have documented **step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation**. My steps: - Start with the initial platform/version/OS and dependencies used, - Specify exact install/launch/configure commands, - List URLs visited, user input (incl. example values/emails/passwords if needed), - Describe all options and toggles enabled or changed, - Include any files or environmental changes, - Identify the expected and actual result at each stage, - Ensure any reasonably skilled user can follow and hit the same issue. ### Expected Behavior Expected behavior: show `reasoning_content` in the frontend folded think-section, and when reasoning_content is over and content starts, finish the think-section and show content outside.. ### Actual Behavior When working on think-enabled models such as qwen3 on latest openAI server deployed with lmdeploy, which returns first `reasoning_content` then `content` **without** `</think>` clouse, the open-webui frontend shows a ever-lasting thinking section including both the reasoning_content and the actual content. lmdeploy's API's behavior ( stream response): ``` ... content=<think>... ... reasoning_content="XX"... ... reasoning_content="XX"... ... reasoning_content="XX"... ... content="XX"... ... content="XX"... ``` vllm's API's behavior ( stream response): ``` ... content=""... ... reasoning_content="XX"... ... reasoning_content="XX"... ... reasoning_content="XX"... ... content="XX"... ... content="XX"... ``` without reasoning enabled: ``` ... content="<think>"... ... content="XX"... ... content="XX"... ... content="</think>"... ... content="XX"... ... content="XX"... ``` Open-webuiworks OK when not enabling reasoning or enabled with vllm, NOT with deploy ### Steps to Reproduce Deploy mode with lmdeploy ```bash lmdeploy serve api_server \ Qwen/Qwen3-32B \ --reasoning-parser qwen-qwq \ --tool-call-parser qwen ``` ### Logs & Screenshots No error in logs. ### Additional Information _No response_
GiteaMirror added the bug label 2025-11-11 16:26:27 -06:00
GiteaMirror commented 2025-11-11 16:26:28 -06:00
Author
Owner
Copy Link
Copy Source

@tjbck commented on GitHub (Jun 24, 2025):

This is a model level issue.

@tjbck commented on GitHub (Jun 24, 2025): This is a model level issue.
GiteaMirror commented 2025-11-11 16:26:28 -06:00
Author
Owner
Copy Link
Copy Source

@jingyibo123 commented on GitHub (Jun 25, 2025):

@tjbck I'd like to know what is OpenWebUI's expected streaming response format for a think-enabled openAI-compatible chat API?

@jingyibo123 commented on GitHub (Jun 25, 2025): @tjbck I'd like to know what is OpenWebUI's expected streaming response format for a think-enabled openAI-compatible chat API?
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
bug
confirmed
confirmed issue
core
documentation
enhancement
good first issue
help wanted
non-core
pull-request
Mirrored from GitHub Pull Request
 python
question
testing wanted
No labels bug
Milestone
No items
No Milestone
Projects
Clear projects
No projects
Assignees
Clear assignees
GiteaMirror ninjasurge (Kaehlen Rog)
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/open-webui#5618