Cannot use all models in Ctrl+K Inline Edits

Describe the Bug

Prior to the update I was able to select any of my “active” models in the inline editing mode (Ctrl+K), now however it just gives me the option to “add more models”, which takes my to my active models page, which shows more active models than are shown in the inline edit.

Steps to Reproduce

  1. Start an inline edit with Ctrl+K
  2. Try to select a specific model out of all you enabled (e.g. Gemini Models don’t work for me here)

Expected Behavior

All active models should be selectable for inline edits.

Screenshots / Screen Recordings

Operating System

Windows 11

Current Cursor Version (Menu → About Cursor → Copy)

Version: 1.0.0 (user setup)
VSCode Version: 1.96.2
Commit: 53b99ce608cba35127ae3a050c1738a959750860
Date: 2025-06-04T19:44:25.253Z
Electron: 34.5.1
Chromium: 132.0.6834.210
Node.js: 20.19.0
V8: 13.2.152.41-electron.0
OS: Windows_NT x64 10.0.26100

Additional Information

(I have all gemini models enabled but none of them show)

Does this stop you from using Cursor

No - Cursor works, but with this issue

2 Likes


A list of all my active models in comparison :slight_smile:

1 Like

I too am experiencing the same isssue.
there is a pattern here. no thinking models are appearing in ctrl + k

1 Like

I have a different but similar issue. I cannot unset “auto” mode and select any model of my choosing.

same issue here.

Same problem happened since the 1.0 update.
Like many I often use for inline cmd+K flash Gemini models via API. It’s impossible to use anything labelled as “thinking” model now, hope they can fix it soon @deanrie @danperks

I believe thinking models have been disabled for CMD+K as the performance of them was very poor from our experience.

Did you see good performance with thinking models in Inline Edit?

@danperks
I really enjoyed gemini-2.5-flash for inline edits as it is extremely fast and cheap at the same time. I generally did not use other thinking models for inline edits to be fair but gemini flash is just such a goated model for such simple tasks.

My workflow with cursor is probably relatively conservative compared to others where I mostly use inline edits and rarely ever the chat with the “more expensive” premium requests and almost never utilize agent mode at all. For that gemini flash was just really great as it balanced capability with speed which is perfect for my use case :slight_smile:

Generally I would appreciate if the user gets the option to decide which models they can use though instead of just removing the option without any notice (maybe grey them out and add a tooltip that says “thinking models often perform worse for inline edits” or sth. instead ^^

3 Likes

I cannot give a valid opinion as I was only using gemini flash for inline stuff or quick question.
I keep other thinking models/requests for real chat/agent and do not waste them for limited context with inline cmd+K.

Similar to @Polarising I never had problems with gemini flash, to be honest it feels like it’s the kind of model suited for inline except now it’s a thinking model.

I understand people might abuse it in cmd+K since it’s free in Cursor, but at least for people with an API key, it would be nice not to be locked out too. If the team could think about it, it’d be appreciated.

Note that Gemini flash unlike Pro can have thinking mode disabled

2 Likes

I think they should come back; they can be useful sometimes in certain cases, and competitors like GitHub Copilot use thinking models on inline edit.