Kimi-K3 looped for 31.5M tokens

Support Bug Reports
1

I was trying Kimi-K3 (low) to review a python function and the agent got stuck in a loop. I didn’t catch it cause I scrolled away to do some other business and came back to see the same message repeated over in the chat. I interrupted to ask if it was stuck and it confirmed. It then completed the simple task. I looked at the dashboard and it used 31.5M tokens in that loop failure (nearly 30% of my allowance).

Am I in the wrong because I assumed the agent wouldn’t need babysitting while I allowed it to work on a task? I’ve seen agents get looped, but not to this extreme before (usually self-correcting after a few passes). I won’t be using Kimi anymore after this, that’s for sure.

Screenshot 2026-09-05 at 1.37.50 PM
Screenshot 2026-09-05 at 1.37.50 PM1690×102 6.18 KB

Screenshot 2026-09-05 at 1.40.54 PM
Screenshot 2026-09-05 at 1.40.54 PM1488×2328 360 KB

2

Hi there!

We detected that this may be a bug report, so we’ve moved your post to the Bug Reports category.

To help us investigate and fix this faster, could you edit your original post to include the details from the template below?

Bug Report Template - Click to expand

Where does the bug appear (feature/product)?

  • Editor, Tab & Chat (autocomplete, Composer, in-editor agent)
  • Terminal & commands
  • Models, pricing & API keys (availability, Auto/Max, BYOK/Bedrock)
  • MCP & tools
  • Cloud Agents & Automations (cursor.com/agents, scheduled/event)
  • BugBot & Code Review
  • Cursor CLI
  • Cursor Mobile
  • Remote (SSH / Dev Containers / WSL)
  • Account, billing & login
  • Something else…

Describe the Bug
A clear and concise description of what the bug is.

Steps to Reproduce
How can you reproduce this bug? We have a much better chance at fixing issues if we can reproduce them!

  1. …
  2. …
  3. …

Expected Behavior
What is meant to happen here that isn’t working correctly?

Screenshots / Screen Recordings
If applicable, attach images or videos (.jpg, .png, .gif, .mp4, .mov)

Operating System

  • Windows 10/11
  • MacOS
  • Linux

Version Information

  • For Cursor IDE: Menu → About Cursor → Copy
  • For Cursor CLI: Run agent about in your terminal
IDE:
Version: 2.xx.x
VSCode Version: 1.105.1
Commit: ......

CLI:
CLI Version 2026.01.17-d239e66

For AI issues: which model did you use?
Model name (e.g., Sonnet 4, Tab…)

For AI issues: add Request ID with privacy disabled
Request ID: f9a7046a-279b-47e5-ab48-6e8dc12daba1
For Background Agent issues, also post the ID: bc-…

Additional Information
Add any other context about the problem here.

Does this stop you from using Cursor?

  • Yes - Cursor is unusable
  • Sometimes - I can sometimes use Cursor
  • No - Cursor works, but with this issue

The more details you provide, the easier it is for us to reproduce and fix the issue. Thanks!

5

Where does the bug appear (feature/product)?

  • Editor, Tab & Chat (autocomplete, Composer, in-editor agent)
  • Terminal & commands
  • Models, pricing & API keys (availability, Auto/Max, BYOK/Bedrock)
  • MCP & tools
  • Cloud Agents & Automations (cursor.com/agents, scheduled/event)
  • BugBot & Code Review
  • Cursor CLI
  • Cursor Mobile
  • Remote (SSH / Dev Containers / WSL)
  • Account, billing & login
  • Something else…

Describe the Bug
A clear and concise description of what the bug is.

Agent got stuck in a loop for over 9 minutes.

Steps to Reproduce
How can you reproduce this bug? We have a much better chance at fixing issues if we can reproduce them!

  1. Asked Kimi-K3-low to do a simple review task and it looped. Not sure how to reproduce.

Expected Behavior
What is meant to happen here that isn’t working correctly?

Agent shouldn’t loop. If it does, it should have a way to break loop instead of continuing indefinitely.

Screenshots / Screen Recordings
If applicable, attach images or videos (.jpg, .png, .gif, .mp4, .mov)
screenshots in original post.

Operating System

  • Windows 10/11
  • MacOS
  • Linux

Version Information

  • For Cursor IDE: Menu → About Cursor → Copy
  • For Cursor CLI: Run agent about in your terminal
Version: 3.18.25
VS Code Extension API: 1.128.0
Commit: 280eca2911f1774689696e5f1efa5a4f97a87af0
Date: 2026-08-31T23:08:02.261Z
Layout: IDE
Build Type: Stable
Release Track: Default
Electron: 40.10.3
Chromium: 144.0.7559.236
Node.js: 24.15.0
V8: 14.4.258.32-electron.0
xterm.js: 6.1.0-beta.291
OS: Darwin x64 22.6.0

For AI issues: which model did you use?
Model name (e.g., Sonnet 4, Tab…) Kimi-K3-low

For AI issues: add Request ID with privacy disabled
Request ID: f9a7046a-279b-47e5-ab48-6e8dc12daba1
For Background Agent issues, also post the ID: bc-…

Additional Information
Add any other context about the problem here.

Does this stop you from using Cursor?

  • Yes - Cursor is unusable
  • Sometimes - I can sometimes use Cursor
  • No - Cursor works, but with this issue
7

Hey @andrewsipe, thanks for the detailed report, and sorry about the wasted usage.

No, you’re not in the wrong here. Expecting an agent to complete a task without constant supervision is totally reasonable. What you described, the agent getting stuck repeating the same step for about ten minutes and not breaking out of the loop on its own, isn’t caused by your setup. This is a known issue with Kimi K3 that we’re tracking. Your dashboard screenshots (31,5M tokens) and the repeated message show it clearly. I can’t share an exact timeline yet, but I’ll post here if there’s an update.

A couple things that can help for now:

  • If you see the same message repeat twice in a row, hit Stop right away and don’t let it keep running.
  • For Kimi K3 specifically, it’s better to run short, focused chats instead of continuing a long session.

If you want us to review this specific session for the usage spent, email [email protected] with a link to this thread and they’ll take a look.

9

Thanks for the response.

I’ll reach out to [email protected] for a review/refund on the token usage. I haven’t experienced this issue with this much token usage before, so that where the frustration truly lies.