Reference
Usage & token quota
The token limits that apply to reviews, how the windows reset, what happens when a limit is reached, and where to see your usage.
Each plan comes with a token budget for the model calls of reviews. You do not buy tokens: the budget is part of the plan and limits how much review work runs in a given period.
Limits
| Plan | 7-day window | 5-hour window | Monthly allowance |
|---|---|---|---|
| Free | 25,000 | 8,000 | 100,000 |
| Pro | 250,000 | 80,000 | 1,000,000 |
| Business | 1,250,000 | 400,000 | 5,000,000 |
| Enterprise | Unlimited | Unlimited | Unlimited |
- The 7-day window is a quarter of the monthly allowance; the 5-hour window is 32% of the 7-day window.
- When a window ends, unused tokens are not carried over. The next window starts at zero.
- The monthly allowance resets when a paid subscription renews.
What happens at a limit
| Limit reached | Result on the pull request |
|---|---|
| 5-hour or 7-day window | The review is paused before any model call. codingassist posts a comment and the status check fails with "Review paused". No tokens are spent |
| Window almost used | The review runs with a shorter AI output. The summary shows a Partial review banner |
| Monthly allowance | The review is skipped. The summary comment is posted with no findings and the status check passes with 0 findings |
A review is never charged for a pull request with more than 150 changed files: it stops before any model call. See Why didn't codingassist comment on my pull request?
See your usage
- The Usage page in the web app shows both windows. It is visible to the Owner.
- Anyone who can comment on a reviewed pull request can reply
@codingassist rate limitto see the plan, the share of the monthly allowance used and the reset date.
Related
Was this page helpful?