[GH-ISSUE #22956] issue: Continue Generation may inject reasoning/tool-call content into the final response body, causing rendering issues

GiteaMirror commented

2026-04-20 02:21:34 -05:00

Owner

Originally created by @ShirasawaSama on GitHub (Mar 23, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/22956

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

Git Clone

Open WebUI Version

v0.8.10

Ollama Version (if applicable)

No response

Operating System

MacOS 26

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

Clicking Continue Generation should continue the answer in a way that preserves correct rendering of the final message.

It should not insert reasoning content, tool-call traces, or other internal generation artifacts into the visible response body. The continued output should remain clean user-facing content, without breaking Markdown, code blocks, or overall message formatting.

Actual Behavior

After the original response stops, clicking Continue Generation may produce output where:

reasoning blocks are inserted into the main response
tool-call/internal generation artifacts appear in the visible content
markdown/code rendering becomes broken or malformed
the final answer contains mixed internal and user-facing text

Steps to Reproduce

Start a conversation with a model that supports reasoning and/or tool calling
Let the assistant generate a long response
Interrupt it, or wait until generation stops unexpectedly
Click Continue Generation
Observe that reasoning/tool-related content may appear inside the final visible response body

Logs & Screenshots

Additional Information

I believe Continue Generation should behave as if OpenWebUI sends a new user message like "continue" (or an equivalent continuation prompt), and then lets the model generate the next assistant turn normally.

It should not continue appending raw generation state from the interrupted response in a way that leaks reasoning/tool-call content into the visible final answer.

Originally created by @ShirasawaSama on GitHub (Mar 23, 2026). Original GitHub issue: https://github.com/open-webui/open-webui/issues/22956 ### 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 Git Clone ### Open WebUI Version v0.8.10 ### Ollama Version (if applicable) _No response_ ### Operating System MacOS 26 ### 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 Clicking Continue Generation should continue the answer in a way that preserves correct rendering of the final message. It should not insert reasoning content, tool-call traces, or other internal generation artifacts into the visible response body. The continued output should remain clean user-facing content, without breaking Markdown, code blocks, or overall message formatting. ### Actual Behavior After the original response stops, clicking Continue Generation may produce output where: - reasoning blocks are inserted into the main response - tool-call/internal generation artifacts appear in the visible content - markdown/code rendering becomes broken or malformed - the final answer contains mixed internal and user-facing text ### Steps to Reproduce - Start a conversation with a model that supports reasoning and/or tool calling - Let the assistant generate a long response - Interrupt it, or wait until generation stops unexpectedly - Click Continue Generation - Observe that reasoning/tool-related content may appear inside the final visible response body ### Logs & Screenshots <img width="2012" height="1802" alt="Image" src="https://github.com/user-attachments/assets/5154e1b5-8c81-400d-a867-3c8e0c36bd7c" /> ### Additional Information I believe Continue Generation should behave as if OpenWebUI sends a new user message like "continue" (or an equivalent continuation prompt), and then lets the model generate the next assistant turn normally. It should not continue appending raw generation state from the interrupted response in a way that leaks reasoning/tool-call content into the visible final answer.