Besides similar problems mentioned below, Cursor had disabled thinking models for inline chat, but we could still use them like Gemini models with our own API key but now even with our own key, they are not visible to select for inline chat (probably because they are merged with the Cursor Gemini models)
Anyway, I would really like to bring this back or at least let us use thinking model for inline chat. Almost all models are thinking/hybrid now. @deanrie had forwarded this request but as usual the team isn’t giving any answers back.
and
Steps to Reproduce
add and activate your Gemini API, add a Gemini flash model
Expected Behavior
seeing the Gemini flash model in the inline chat list
Operating System
Windows 10/11
Current Cursor Version (Menu → About Cursor → Copy)
This is a known limitation - the team intentionally disabled thinking models in inline edit CMD+K due to poor performance. This applies to all thinking models, including Gemini Flash with BYOK.
The team is considering adding model selection for inline edit, but there’s no specific solution or ETA yet.
That’s encouraging news @deanrie that you guys are considering it and better align with the competitions. Please keep us updated.
I’m not a fan of wasting Cursor credits on non-thinking Claude for simple inline questions and don’t want to clutter my models list of non thinking models.
If you don’t want to globally enable Cursor thinking models with inline chat, with BYOK at least, we could use Gemini Flash, it was still working pre 2.0.
Never had performance issues nor saw complaints about inline chat.
As the term imply, it’s mainly for quick questions and simple things, staying focused in the editor, without having to waste time open a whole new chat+having to delete etc…
