Ability to turn off reasoning/thinkinh

[Geo-Vermont]

What problem would this solve?

I'm not an expert with local LLM's, I want to be straightforward.
It's more convenient to use an llm without reasoning , For eg: When I use Ollama through the terminal I can simply use "Ollama run qwen3.6-35b:latest --think=false" to turn off thinking when not needed.

It would be convenient to add a toggle to turn off reasoning.
Thanks.

Proposed behavior

A dedicated toggle either in the settings menu or the direct chat to turn off reasoning

Alternatives considered

N/A

Additional context

N/A

Before submitting

• I searched the existing issues and did not find the same request.

[engeldlgado]

This already exists for the in-app chat: open the parameters button (sliders icon) next to the message box and set Reasoning effort to Off. That sends enable_thinking: false plus /no_think, the same effect as --think=false in Ollama. It's disabled only when the loaded model's chat template has no thinking support, and reasoning-only models (e.g. R1) may still think regardless.

One question: do you need it for the app's chat, or for an external client talking to the server? The chat toggle is per-conversation; for a server-wide default there's currently no switch (you'd have to add --reasoning-budget 0 under Settings → Extra arguments). If that's your case, I'll add a proper toggle.