Agent was stocked in thinking very slowly

Support Bug Reports
,
1

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

work can’t finish correctly.
After doing for a few minutes,the agent was stocked in thinking.It was’t stopped,look like still in thinking,but a few minutes later,thinking was not finished.Look like thinking very slowly.
Stop the seesion by manual,redo in new seesion.It was the same.

Steps to Reproduce

Stop the seesion by manual,redo in new seesion.It was the same.

Operating System

Windows 10/11

Version Information

Version: 3.19.7 (user setup)
VS Code Extension API: 1.128.0
Commit: 90de2327392570a5f5f625c656c6749d228e6430
Date: 2026-09-02T23:03:07.739Z
Layout: Agent Window
Build Type: Stable
Release Track: Default
Electron: 42.10.0
Chromium: 148.0.7778.280
Node.js: 24.18.1
V8: 14.8.178.38-electron.0
xterm.js: 6.1.0-beta.291
OS: Windows_NT x64 10.0.26200

For AI issues: which model did you use?

Grok 4.6

For AI issues: add Request ID with privacy disabled

0af667cb-0a9b-4555-ba96-cbd8ffc6961e

Does this stop you from using Cursor

Sometimes - I can sometimes use Cursor

7

Hey @sm_s thanks for the report, and for the request ID. This isn’t anything wrong with your setup, and it’s an issue we’re tracking.

A few things that should unblock you in the meantime:

  1. If a turn keeps “thinking” for more than a couple of minutes with nothing happening, hit Stop rather than waiting it out.
  2. Start a fresh chat and keep each request small and specific. That conversation had grown quite large, which makes this much more likely.
  3. For that step, switch to Auto or another model. The same task usually goes through fine.

If it still happens in a fresh chat, reply with the new request ID and I’ll take another look.

10

Hey @sm_s,

Following up on this one. We found what was going on with that request: Grok 4.6 was getting stuck repeating itself while reasoning, and Cursor kept retrying the same step instead of catching it. That is what showed up as thinking that never finished.

This has now been addressed on our side and will be rolled out soon if it hasn’t been already. It doesn’t need a Cursor update, it applies to your agents automatically.

Could you give Grok another try on the kind of task that was getting stuck and let me know if the agent still stalls in thinking? If it does, a fresh request ID would help a lot.