According to the Cursor updated models page Grok-3-Beta should now be capable of using it’s thinking mode, all the other models seem to be updated but when you view the Grok-3-Beta model in the model’s dropdown list in the IDE it still has no thinking indicator and when used performs no reasoning ever.
As you can see in the screenshot of the models list is should have thinking enabled:
As you can see in Cursor thinking is missing still:
My current Cursor version:
Version: 0.50.3
VSCode Version: 1.96.2
Commit: 7ae22cf8cd5af9e08b62585dd03d10f5f610acf0
Date: 2025-05-13T02:19:07.844Z
Electron: 34.3.4
Chromium: 132.0.6834.210
Node.js: 20.18.3
V8: 13.2.152.41-electron.0
OS: Linux x64 6.12.10-76061203-generic
The issue doesn’t prevent the use of Cursor but does significantly kneecap the performance of the Grok-3-Beta model.
I do not see this being helpful but here is a Request ID of a detailed prompt in a new chat that displays thinking not working despite the fact the prompt was complex - c3bf9294-f228-4ea3-8346-2d128f874384