Copilot usage limit reached: what stops, what keeps working
Chat and agents pause, completions carry on, and you can wait, pay or upgrade until the monthly reset.
By ClaudeGPT Si Newsroom · Editorial team
· 3 min read
Hitting your Copilot usage limit means you have spent your plan's monthly AI credits, so chat, the CLI and agents pause until you pay for more, upgrade, or reach the reset at 00:00 UTC on the first of the month. Code completions keep working on paid plans. To see the limit coming instead of finding it mid-task, keep a tracker such as UsageAtlas open on your desktop.
Why the limit feels different now
On June 1, 2026, GitHub moved every monthly Copilot plan from premium requests to AI credits. Credits are spent on tokens, at the API price of the model you choose. Long agent runs and large models drain the balance faster than short questions, so two developers on the same plan can reach the limit weeks apart.
GitHub also dropped the old safety net. Under premium requests, running out could drop you to a cheaper model. That fallback no longer exists.
What still works and what stops
| Feature | After you hit the limit |
|---|---|
| Code completions | Keep working on paid plans |
| Next edit suggestions | Keep working on paid plans |
| Copilot Chat | Paused until you add usage or the month resets |
| Copilot CLI and agents | Paused until you add usage or the month resets |
| Spaces and Spark | Paused until you add usage or the month resets |
Your three options
- Wait. The allowance refills at 00:00 UTC on the first day of each calendar month. Unused credits never carry over, so there is nothing saved from last month.
- Pay for more. You can set a dollar budget for additional usage. GitHub warns that extra usage may be capped, and you may need to pay what you have already used before continuing.
- Upgrade. Moving to a higher plan gives you its larger allowance, less what you used this month.
If your seat comes from your employer, the choice is your admin's. Organizations decide whether to allow usage past the pool or cap it.
Where to check how close you are
- In the editor: click the Copilot icon in the status bar in VS Code, JetBrains or Eclipse.
- On GitHub: open your billing settings and filter Metered usage to Copilot.
- On your desktop: UsageAtlas, free and open source, shows the quotas and seat credits GitHub reports, with the reset time when it is supplied. Set a threshold and it sends one alert per reset cycle, so you get a heads-up at, say, most of the way through rather than at zero.
UsageAtlas runs on Windows, macOS and Linux and also covers Claude Code, Codex and Cursor, which helps if you switch tools when Copilot runs low. It never invents a limit: if GitHub sends no allowance, the limit shows as unknown and only your usage is counted. For the exact figure you will be billed, GitHub's own page is the one to trust.
On an annual plan?
Annual Pro and Pro+ subscribers still count premium requests, with a multiplier per model, until the plan expires. Their counter also resets on the first of each month at 00:00 UTC.
FAQ
What happens when I reach my Copilot usage limit? Chat, the CLI and agents stop until you add usage, upgrade, or the monthly reset. Completions continue on paid plans.
When does the Copilot usage limit reset? At 00:00 UTC on the first day of each calendar month.
Why did I hit my Copilot limit so early? Credits follow tokens. Long agent sessions and larger models spend them faster; GitHub's usage CSV shows which models you used.