GPT-5.4 High model missing on Cursor Desktop for Mac

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

The GPT-5.4 High model is no longer appearing in the model selection list in the Cursor IDE desktop app for macOS.

This model was previously available and working normally, but it has disappeared from the available models list. Other models are still visible and selectable.

I have already restarted the application, but the issue persists.

Could you please verify whether this model has been removed, is temporarily unavailable, or if this is a bug affecting the macOS desktop version?

Steps to Reproduce

  1. Open Cursor IDE on macOS
  2. Open any chat, composer, or AI assistant panel
  3. Click the model selector dropdown
  4. Check the list of available models
  5. Observe that “GPT-5.4 High” is missing from the list

Expected Behavior

The GPT-5.4 High model should appear in the model selection dropdown as it did previously and be available for use.

Screenshots / Screen Recordings

Operating System

MacOS

Version Information

Cursor Version: 3.1.15 (Universal)
VSCode Version: 1.105.1
Commit: 3a67af7b780e0bfc8d32aefa96b8ff1cb8817f80
Date: 2026-04-15T01:46:06.515Z
Layout: editor
Build Type: Stable
Release Track: Default
Electron: 39.8.1
Chromium: 142.0.7444.265
Node.js: 22.22.1
V8: 14.2.231.22-electron.0
OS: Darwin arm64 25.3.0

For AI issues: which model did you use?

Previously used model: GPT-5.4 High
Current issue: model no longer available in selector

For AI issues: add Request ID with privacy disabled

N/A - issue occurs before sending any request because the model is missing from the selector

Additional Information

The issue started recently and appears to affect only the desktop version on macOS.

I am using Apple Silicon (arm64).

Other available models still appear normally, and the problem seems specific to GPT-5.4 High.

Does this stop you from using Cursor

No - Cursor works, but with this issue

To be honest, the interaction function of switching AI models seems not very user-friendly in the new version. From version 3.1.15.

I think this actually kind of makes more sense, especially since it seems like it remembers your selections.

This seems like a better UI than having 10–20 “models” that all have slight variations: GPT-5.4 (low) fast, GPT-5.4 (low) non-fast, GPT-5.4 (low) fast 1M context, GPT-5.4 (low)—and on and on.

Hey, this isn’t a bug. In the latest versions, we updated the UI for the model picker. Instead of separate rows for each option (GPT-5.4, GPT-5.4 High, GPT-5.4 Extra High, etc.), there’s now one model, and you set the reasoning level separately.

To pick GPT-5.4 High:

  1. Select GPT-5.4 in the model picker
  2. Hover over the model name and click the edit button
  3. Switch reasoning to High

More details with a screenshot from our team: What happend to GTP-5.4-Extra-High non MAX

Let me know if you can’t see the edit button.

Thank you for the clarification and for helping me access the model again.

@deanrie I like the improvement but the default of the model should be based on what we were using not the medium reasoning option. If I’m using most of the time gpt 5.4 high before the update, after the update it should default to that model not the medium reasoning effort. It was a shock seeing the change, choosing the model and reasoning effort is something very sensitive, and while in shock I did not even noticed the edit button.

Hey @jsandlerus, the feedback is valid. After the migration, the default should keep the previously used reasoning level, not reset to medium.

For reference, the reasoning level is now saved per chat, so once you set it to High, it stays that way.

This is a major, non-obvious UI change, with zero notification or anything in the changelog - please don’t do that. Was this from one patch version rather than a minor or major? If so, that makes it even worse.

The change itself is poor for those of us who frequently switch between just two thinking modes (does anyone really use “low”?). Instead of 2 interactions it’s now 4 and much more fiddly, with a tiny edit button.

Right! I just updated Cursor from version 2.5 to the newest version, and I noticed the issue described here. I thought it was a bug at first. I already regret updating—it’s a disaster having to click so many times just to switch between 5.5 and 5.5 Extra High.

Hey @Buck3thead, we hear your feedback. A couple things that make using the new picker a lot easier:

  • Cmd/Ctrl+Shift+/ cycles the reasoning effort right in chat, so you don’t need to open the picker and hit the edit button.
  • The reasoning level is saved per chat, so you set it once and leave it until you want to change it.

On the overall UX feedback, the click count for frequent switching went up. I’ll pass that to the team, thanks for explaining it clearly.

New model picker definitely helps with the long lists, but I second the UX feedback that switching between different options/context/efforts for one specific model is a bit harder now. I usually start some tasks with high (or some other specific “setting”) and move to medium later. I guess you can’t make everyone happy. :slight_smile:

Something like a “favorite” or “most used” option could be useful to allow quick switching.