I have searched the existing issues and discussions.
I am using the latest version of Open WebUI.
Installation Method
Docker
Open WebUI Version
v0.6.15
Ollama Version (if applicable)
No response
Operating System
Ubuntu 22.04
Browser (if applicable)
Chrome 137.0.7151.122 64bit
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
.
Actual Behavior
.
Steps to Reproduce
Navigate to the chat interface.
Select any model (the issue was observed with Gemini-2.0-flash but may affect others).
Enter the following prompt that triggers the issue:
self.prefix = "<|im_start|>system\nJudge whether the Document meets the requirements based on the Query and the Instruct provided. Note that the answer can only be \"yes\" or \"no\".<|im_end|>\n<|im_start|>user\n"
self.suffix = "<|im_end|>\n<|im_start|>assistant\n<think>\n\n</think>\n\n"
Submit the prompt and wait for the LLM to respond.
Observe the UI as the response is being streamed/rendered.
Logs & Screenshots
Additional Information
No response
Originally created by @smn3786 on GitHub (Jul 2, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/15461
### Check Existing Issues
- [x] I have searched the existing issues and discussions.
- [x] I am using the latest version of Open WebUI.
### Installation Method
Docker
### Open WebUI Version
v0.6.15
### Ollama Version (if applicable)
_No response_
### Operating System
Ubuntu 22.04
### Browser (if applicable)
Chrome 137.0.7151.122 64bit
### 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
.
### Actual Behavior
.
### Steps to Reproduce
1. Navigate to the chat interface.
2. Select any model (the issue was observed with `Gemini-2.0-flash` but may affect others).
3. Enter the following prompt that triggers the issue:
```
self.prefix = "<|im_start|>system\nJudge whether the Document meets the requirements based on the Query and the Instruct provided. Note that the answer can only be \"yes\" or \"no\".<|im_end|>\n<|im_start|>user\n"
self.suffix = "<|im_end|>\n<|im_start|>assistant\n<think>\n\n</think>\n\n"
```
4. Submit the prompt and wait for the LLM to respond.
5. Observe the UI as the response is being streamed/rendered.
### Logs & Screenshots
### Additional Information
_No response_
GiteaMirror
added the bug label 2026-04-25 06:57:56 -05:00
I haven't seen anything strange in streaming or rendering, maybe a model issue.
<!-- gh-comment-id:3066289683 -->
@rgaricano commented on GitHub (Jul 13, 2025):
I haven't seen anything strange in streaming or rendering, maybe a model issue.
I'm not entirely sure if it's a line break literal (\n) or a pipe symbol (|), but I think the rendering process might need some fixes in the escaping handling.
Try making the model repeat the following without using a code block:
self.prefix = "<|im_start|>system\nJudge whether the Document meets the requirements based on the Query and the Instruct provided. Note that the answer can only be \"yes\" or \"no\".<|im_end|>\n<|im_start|>user\n"
self.suffix = "<|im_end|>\n<|im_start|>assistant\n<think>\n\n</think>\n\n"
<!-- gh-comment-id:3082298887 -->
@smn3786 commented on GitHub (Jul 17, 2025):
<img width="684" height="489" alt="Image" src="https://github.com/user-attachments/assets/37a889dd-8539-44da-b284-4c4b466d9fa8" />
I'm not entirely sure if it's a line break literal (\n) or a pipe symbol (|), but I think the rendering process might need some fixes in the escaping handling.
Try making the model repeat the following without using a code block:
```
self.prefix = "<|im_start|>system\nJudge whether the Document meets the requirements based on the Query and the Instruct provided. Note that the answer can only be \"yes\" or \"no\".<|im_end|>\n<|im_start|>user\n"
self.suffix = "<|im_end|>\n<|im_start|>assistant\n<think>\n\n</think>\n\n"
```
Ahhhh I think the think tag was the issue.
I'm not sure if any other characters are messing up the rendering, but it completely breaks when I use the Gemini model.
<!-- gh-comment-id:3082313446 -->
@smn3786 commented on GitHub (Jul 17, 2025):
<img width="738" height="1104" alt="Image" src="https://github.com/user-attachments/assets/a2aad82b-945e-4f30-850d-5008a035f2fe" />
Ahhhh I think the `think tag` was the issue.
I'm not sure if any other characters are messing up the rendering, but it completely breaks when I use the Gemini model.
Ok, <think> tag doesn't seem like a big issue.
And if the rendering problem only happens with Google models, then yeah, it's probably the model's issue like you said.
So I'll close this issue.
I'm really happy that there's such a great open-source F/E app like this! Hope it keeps improving in the future!
<!-- gh-comment-id:3086040907 -->
@smn3786 commented on GitHub (Jul 18, 2025):
Ok, `<think>` tag doesn't seem like a big issue.
And if the rendering problem only happens with Google models, then yeah, it's probably the model's issue like you said.
So I'll close this issue.
I'm really happy that there's such a great open-source F/E app like this! Hope it keeps improving in the future!
Rendering looks weird when the response from the model contains <think> .
<!-- gh-comment-id:3211347678 -->
@alanxmay commented on GitHub (Aug 21, 2025):
Same issue.
Rendering looks weird when the response from the model contains `<think>` .
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Originally created by @smn3786 on GitHub (Jul 2, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/15461
Check Existing Issues
Installation Method
Docker
Open WebUI Version
v0.6.15
Ollama Version (if applicable)
No response
Operating System
Ubuntu 22.04
Browser (if applicable)
Chrome 137.0.7151.122 64bit
Confirmation
README.md.Expected Behavior
.
Actual Behavior
.
Steps to Reproduce
Navigate to the chat interface.
Select any model (the issue was observed with
Gemini-2.0-flashbut may affect others).Enter the following prompt that triggers the issue:
Submit the prompt and wait for the LLM to respond.
Observe the UI as the response is being streamed/rendered.
Logs & Screenshots
Additional Information
No response
@rgaricano commented on GitHub (Jul 13, 2025):
I haven't seen anything strange in streaming or rendering, maybe a model issue.
@smn3786 commented on GitHub (Jul 17, 2025):
I'm not entirely sure if it's a line break literal (\n) or a pipe symbol (|), but I think the rendering process might need some fixes in the escaping handling.
Try making the model repeat the following without using a code block:
@smn3786 commented on GitHub (Jul 17, 2025):
Ahhhh I think the
think tagwas the issue.I'm not sure if any other characters are messing up the rendering, but it completely breaks when I use the Gemini model.
@rgaricano commented on GitHub (Jul 17, 2025):
@smn3786 commented on GitHub (Jul 18, 2025):
Ok,
<think>tag doesn't seem like a big issue.And if the rendering problem only happens with Google models, then yeah, it's probably the model's issue like you said.
So I'll close this issue.
I'm really happy that there's such a great open-source F/E app like this! Hope it keeps improving in the future!
@alanxmay commented on GitHub (Aug 21, 2025):
Same issue.
Rendering looks weird when the response from the model contains
<think>.