AI coding tool usage-limit resets, in numbers
Every usage-limit reset announced by an AI coding tool's company or its staff, counted the same way for every tool. Updated Oct 29, 2026.
Codex
Codex has reset its usage limits 64 times since Sep 17, 2025.
- 28 resets in the last 90 days, a median of 2d 4h apart.
- Longest wait: 67.7 days, Dec 25, 2025 → Mar 3, 2026.
- None so far between 09:00 and 15:00 UTC.
- Last reset: Oct 28, 2026.
Claude Code
Claude Code has reset its usage limits 14 times since Apr 16, 2026.
- 3 resets in the last 90 days, a median of 10d 11h apart.
- Longest wait: 47.6 days, Jul 16, 2026 → Sep 1, 2026.
- None so far between 04:00 and 16:00 UTC.
- Last reset: Sep 22, 2026.
No announced global reset recorded for Antigravity, Cursor, Devin Desktop, GitHub Copilot, and Kiro.
Every tool
| Tool | Resets | 90 days | 30 days | Median gap (90 days) | Longest wait | Last reset |
|---|---|---|---|---|---|---|
| Codex | 64 | 28 | 13 | 2d 4h | 67.7 days | Oct 28, 2026 |
| Claude Code | 14 | 3 | 0 | 10d 11h | 47.6 days | Sep 22, 2026 |
| Antigravity | 0 | 0 | 0 | — | — | — |
| Cursor | 0 | 0 | 0 | — | — | — |
| Devin Desktop | 0 | 0 | 0 | — | — | — |
| GitHub Copilot | 0 | 0 | 0 | — | — | — |
| Kiro | 0 | 0 | 0 | — | — | — |
Resets per month
Codex
Show the numbers
| Month | Resets |
|---|---|
| Sep 25 | 1 |
| Oct 25 | 0 |
| Nov 25 | 2 |
| Dec 25 | 4 |
| Jan 26 | 0 |
| Feb 26 | 0 |
| Mar 26 | 6 |
| Apr 26 | 6 |
| May 26 | 3 |
| Jun 26 | 5 |
| Jul 26 | 9 |
| Aug 26 | 9 |
| Sep 26 | 6 |
| Oct 26 | 13 |
Claude Code
Show the numbers
| Month | Resets |
|---|---|
| Sep 25 | 0 |
| Oct 25 | 0 |
| Nov 25 | 0 |
| Dec 25 | 0 |
| Jan 26 | 0 |
| Feb 26 | 0 |
| Mar 26 | 0 |
| Apr 26 | 2 |
| May 26 | 1 |
| Jun 26 | 5 |
| Jul 26 | 3 |
| Aug 26 | 0 |
| Sep 26 | 3 |
| Oct 26 | 0 |
Resets by hour of day (UTC)
Codex
Show the numbers
| Hour (UTC) | Resets |
|---|---|
| 00:00 | 7 |
| 01:00 | 3 |
| 02:00 | 5 |
| 03:00 | 6 |
| 04:00 | 6 |
| 05:00 | 4 |
| 06:00 | 2 |
| 07:00 | 1 |
| 08:00 | 2 |
| 09:00 | 0 |
| 10:00 | 0 |
| 11:00 | 0 |
| 12:00 | 0 |
| 13:00 | 0 |
| 14:00 | 0 |
| 15:00 | 1 |
| 16:00 | 3 |
| 17:00 | 4 |
| 18:00 | 2 |
| 19:00 | 3 |
| 20:00 | 4 |
| 21:00 | 3 |
| 22:00 | 2 |
| 23:00 | 6 |
Claude Code
Show the numbers
| Hour (UTC) | Resets |
|---|---|
| 00:00 | 1 |
| 01:00 | 0 |
| 02:00 | 2 |
| 03:00 | 1 |
| 04:00 | 0 |
| 05:00 | 0 |
| 06:00 | 0 |
| 07:00 | 0 |
| 08:00 | 0 |
| 09:00 | 0 |
| 10:00 | 0 |
| 11:00 | 0 |
| 12:00 | 0 |
| 13:00 | 0 |
| 14:00 | 0 |
| 15:00 | 0 |
| 16:00 | 1 |
| 17:00 | 2 |
| 18:00 | 3 |
| 19:00 | 0 |
| 20:00 | 2 |
| 21:00 | 2 |
| 22:00 | 0 |
| 23:00 | 0 |
How it's counted
A reset counts when the company or its staff announce it (for Codex, mostly Tibo's posts on X), and each entry links to its source. Announcements less than an hour apart (Plus and Pro minutes apart, say) are one reset. Retracted entries are left out. Fixed refills (a 5-hour window, a monthly plan) are not resets. A few entries came from a word-for-word mirror of the account: they link to the author's profile, not to the post.
Download the data
CSVJSONCheck it from your own tools: API and MCP server
Free to reuse under CC BY 4.0: please link to this page.
How to cite
QuotaTick, “AI coding tool usage-limit resets”, https://quotatick.com/en/reports/resets (retrieved Oct 29, 2026).