Formatting issue with deepseek-r1

[viktorChekhovoi]

Describe the bug
When using deepseek-r1, the llm "reasoning" has the same formatting as its intended main output, which creates confusion for the user.

To Reproduce
Steps to reproduce the behavior:

1. In a new installation, select "Attach Local LLM"
2. Type 'deepseek-r1' into the text box, download the model and run the app
3. Ask any question in the chat bar

Expected behavior
Similar to DeepSeek's UI, the reasoning should be formatted separately from the rest of the output, with a collapsible area or a smaller/fainter font.

Screenshots

Desktop (please complete the following information):

• OS: Windows

Additional context
I filed this under bugs as opposed to a feature request because, in my opinion, it negatively impacts the user experience.

[weilirs]

Thank you for reporting this bug! It would be even better if you want to fix it😊