Stopping mid-chat

[erlebach]

Describe the Bug

I feel that when asking a question, that once the LLM starts to answer, the ansewr should run to completion. The chatbot should never freeze (at worst, a message should be printed to the user stating what’s going on.

Steps to Reproduce

I doubt this can be reproduced since it is strongly dependent on token usage, context, OS, etc.

Screenshots / Screen Recordings

Operating System

MacOS

Current Cursor Version (Menu → About Cursor → Copy)

Version: 1.3.8
VSCode Version: 1.99.3
Commit: a1fa6fc7d2c2f520293aad84aaa38d091dee6fe0
Date: 2025-07-31T21:33:51.093Z (1 day ago)
Electron: 34.5.1
Chromium: 132.0.6834.210
Node.js: 20.19.0
V8: 13.2.152.41-electron.0
OS: Darwin arm64 24.5.0

Does this stop you from using Cursor

No - Cursor works, but with this issue

[deanrie]

Hey, could you let me know which model you are using? Also, does this happen in a new chat?

[erlebach]

Hi, I am using “auto”. It is possible that this happens once I have exceeded allocation (which I am not warned about). I quit Cursor and get back in, and can get another 1-2 messages, and then it happens again. I then changed to Sonnet 4, and the problem keeps occurring. I suspect the issue would happen with most models. Never in thinking or Max mode.

[deanrie]

Is all this happening in a new chat? Are you getting any errors, or does the generation just get stuck?