[GH-ISSUE #21832] issue: Post-tool “thinking” text leaks outside reasoning tags again #35117

Closed
opened 2026-04-25 09:19:29 -05:00 by GiteaMirror · 4 comments
Owner

Originally created by @kksaohun on GitHub (Feb 24, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/21832

Check Existing Issues

  • I have searched for any existing and/or related issues.
  • I have searched for any existing and/or related discussions.
  • I have also searched in the CLOSED issues AND CLOSED discussions and found no related items (your issue might already be addressed on the development branch!).
  • I am using the latest version of Open WebUI.

Installation Method

Docker

Open WebUI Version

v0.8.5 (latest)

Ollama Version (if applicable)

Operating System

Linux

Browser (if applicable)

Edge

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

Thinking shoule be hidden under "Thought for ....".

Actual Behavior

Image

The LLM is MiniMax M2.5 on VLLM, via OpenAI-compatible API. OpenWebUI's "native" tool calling.

(It looks incredibly similar to #16973, which is older and closed. Maybe it's behaviour that happens again in certain circumstances?)

Steps to Reproduce

  1. Install OpenWebUI
  2. Install vllm + minimax 2.5 (4xGPU)
  3. Switch on "native" tool (function) calling.
  4. Ask something that uses tools. Issue appears about 80% of the time.

Logs & Screenshots

Browser (does not seem relevant):

Image

docker (does not seem relevant):

Image

Additional Information

No response

Originally created by @kksaohun on GitHub (Feb 24, 2026). Original GitHub issue: https://github.com/open-webui/open-webui/issues/21832 ### Check Existing Issues - [x] I have searched for any existing and/or related issues. - [x] I have searched for any existing and/or related discussions. - [x] I have also searched in the CLOSED issues AND CLOSED discussions and found no related items (your issue might already be addressed on the development branch!). - [x] I am using the latest version of Open WebUI. ### Installation Method Docker ### Open WebUI Version v0.8.5 (latest) ### Ollama Version (if applicable) - ### Operating System Linux ### Browser (if applicable) Edge ### 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 Thinking shoule be hidden under _"Thought for ...."_. ### Actual Behavior <img width="1446" height="817" alt="Image" src="https://github.com/user-attachments/assets/8ee92d6b-d500-404d-bf30-bec4a86a0362" /> The LLM is **MiniMax M2.5** on **VLLM**, via OpenAI-compatible API. OpenWebUI's **"native" tool calling**. (It looks incredibly similar to #16973, which is older and closed. Maybe it's behaviour that happens again in certain circumstances?) ### Steps to Reproduce 1. Install OpenWebUI 2. Install vllm + minimax 2.5 (4xGPU) 3. Switch on "native" tool (function) calling. 4. Ask something that uses tools. Issue appears about 80% of the time. ### Logs & Screenshots Browser (does not seem relevant): <img width="2219" height="205" alt="Image" src="https://github.com/user-attachments/assets/e269a733-5659-4e08-82b8-d96e90c635a3" /> docker (does not seem relevant): <img width="2215" height="167" alt="Image" src="https://github.com/user-attachments/assets/8a10cfd1-a543-4ddb-9a0c-ece6374f6007" /> ### Additional Information _No response_
GiteaMirror added the bug label 2026-04-25 09:19:29 -05:00
Author
Owner

@Classic298 commented on GitHub (Feb 24, 2026):

I cant reproduce with hosted MiniMax. Works just fine. Could this be an issue with vLLM?

<!-- gh-comment-id:3954009491 --> @Classic298 commented on GitHub (Feb 24, 2026): I cant reproduce with hosted MiniMax. Works just fine. Could this be an issue with vLLM?
Author
Owner

@kksaohun commented on GitHub (Feb 24, 2026):

Not sure. VLLM is being run with the "official" recommended parameters from here: https://docs.vllm.ai/projects/recipes/en/latest/MiniMax/MiniMax-M2.html#launching-m25m21m2-with-vllm

vllm serve MiniMaxAI/MiniMax-M2.5 \
  --tensor-parallel-size 4 \
  --tool-call-parser minimax_m2 \
  --reasoning-parser minimax_m2_append_think  \
  --enable-auto-tool-choice
<!-- gh-comment-id:3954084671 --> @kksaohun commented on GitHub (Feb 24, 2026): Not sure. VLLM is being run with the "official" recommended parameters from here: <https://docs.vllm.ai/projects/recipes/en/latest/MiniMax/MiniMax-M2.html#launching-m25m21m2-with-vllm> ``` vllm serve MiniMaxAI/MiniMax-M2.5 \ --tensor-parallel-size 4 \ --tool-call-parser minimax_m2 \ --reasoning-parser minimax_m2_append_think \ --enable-auto-tool-choice ```
Author
Owner

@kksaohun commented on GitHub (Feb 24, 2026):

Could be an incompatibility between vllm's --reasoning-parser minimax_m2_append_think and OpenWebUI...?

<!-- gh-comment-id:3954269505 --> @kksaohun commented on GitHub (Feb 24, 2026): Could be an incompatibility between vllm's `--reasoning-parser minimax_m2_append_think` and OpenWebUI...?
Author
Owner

@tjbck commented on GitHub (Feb 24, 2026):

Likely an inference side issue.

<!-- gh-comment-id:3954424998 --> @tjbck commented on GitHub (Feb 24, 2026): Likely an inference side issue.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/open-webui#35117