Originally created by @washcontacnet1 on GitHub (May 15, 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
v0.6.9
Ollama Version (if applicable)
No response
Operating System
Windows 11
Browser (if applicable)
Chrome, Edge, Firefox
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 listed steps to reproduce the bug in detail.
Expected Behavior
Same response in Openwebui and official chat interface (chatgpt or gemini)
Actual Behavior
Hi, I would like to know if anyone else has this problem. I have done thousands of tests and have not been able to reach any conclusion.
When I go directly to chatgpt or Gemini, and ask a sequence of 5 questions, it answers all 5 correctly.
When I ask the same questions via Openwebui, it gets at least 4 wrong.
I have tried using the model directly via API, I have used an assistant created by the platform (https://platform.openai.com/assistants) and incorporated it into Openwebui, the result is the same.
Is anyone else having this scenario?
Steps to Reproduce
Go to openwebui
Send a message (complex question)
Receive the incorrect anwser
Logs & Screenshots
Wrong answer!
Wrong answer!
Correct answer
Correct answer
Additional Information
No response
Originally created by @washcontacnet1 on GitHub (May 15, 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
v0.6.9
### Ollama Version (if applicable)
_No response_
### Operating System
Windows 11
### Browser (if applicable)
Chrome, Edge, Firefox
### 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 listed steps to reproduce the bug in detail.
### Expected Behavior
Same response in Openwebui and official chat interface (chatgpt or gemini)
### Actual Behavior
Hi, I would like to know if anyone else has this problem. I have done thousands of tests and have not been able to reach any conclusion.
When I go directly to chatgpt or Gemini, and ask a sequence of 5 questions, it answers all 5 correctly.
When I ask the same questions via Openwebui, it gets at least 4 wrong.
I have tried using the model directly via API, I have used an assistant created by the platform (https://platform.openai.com/assistants) and incorporated it into Openwebui, the result is the same.
Is anyone else having this scenario?
### Steps to Reproduce
1. Go to openwebui
2. Send a message (complex question)
3. Receive the incorrect anwser
### Logs & Screenshots
Wrong answer!

Wrong answer!

Correct answer

Correct answer

### Additional Information
_No response_
GiteaMirror
added the bug label 2025-11-11 16:13:56 -06:00
Perhaps they have other temperature, top_k etc. settings? Have you tried lower temperature settings than the default 0.8?
@SpizzyCoder commented on GitHub (May 15, 2025):
Perhaps they have other temperature, top_k etc. settings? Have you tried lower temperature settings than the default 0.8?
@washcontacnet1 commented on GitHub (May 15, 2025):
Perhaps they have other temperature, top_k etc. settings? Have you tried lower temperature settings than the default 0.8?
Thanks for your reply.
Yeah, I already tried to lower the temperature.
The results is the same.
@washcontacnet1 commented on GitHub (May 15, 2025):
> Perhaps they have other temperature, top_k etc. settings? Have you tried lower temperature settings than the default 0.8?
Thanks for your reply.
Yeah, I already tried to lower the temperature.
The results is the same.
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 @washcontacnet1 on GitHub (May 15, 2025).
Check Existing Issues
Installation Method
Git Clone
Open WebUI Version
v0.6.9
Ollama Version (if applicable)
No response
Operating System
Windows 11
Browser (if applicable)
Chrome, Edge, Firefox
Confirmation
README.md.Expected Behavior
Same response in Openwebui and official chat interface (chatgpt or gemini)
Actual Behavior
Hi, I would like to know if anyone else has this problem. I have done thousands of tests and have not been able to reach any conclusion.
When I go directly to chatgpt or Gemini, and ask a sequence of 5 questions, it answers all 5 correctly.
When I ask the same questions via Openwebui, it gets at least 4 wrong.
I have tried using the model directly via API, I have used an assistant created by the platform (https://platform.openai.com/assistants) and incorporated it into Openwebui, the result is the same.
Is anyone else having this scenario?
Steps to Reproduce
Logs & Screenshots
Wrong answer!


Wrong answer!
Correct answer

Correct answer
Additional Information
No response
@SpizzyCoder commented on GitHub (May 15, 2025):
Perhaps they have other temperature, top_k etc. settings? Have you tried lower temperature settings than the default 0.8?
@washcontacnet1 commented on GitHub (May 15, 2025):
Thanks for your reply.
Yeah, I already tried to lower the temperature.
The results is the same.