June 8, 2026. The cost of AI coding tools just became a variable line item. On June 1, GitHub switched every Copilot plan to usage-based billing, retiring the old premium request model in favor of GitHub AI Credits. For any team that leans on Copilot to ship faster, the change rewrites how you budget for it.
What changed
- Every plan now bills on AI Credits consumed. Each plan still includes a monthly allotment, and one AI Credit is worth one cent. Usage is calculated from token consumption, counting input, output, and cached tokens at each model's published API rates, so a heavier model burns credits faster.
- Code completions stay unlimited. Inline code completions and next edit suggestions remain uncapped on paid plans. Everything else, including chat and agent-style work, draws down credits.
- Code review now also spends Actions minutes. Copilot code review consumes GitHub Actions minutes in addition to AI Credits, and admins can now set a default Actions runner across all repositories at once.
- User-level budgets are generally available. Organizations and enterprises can set a universal per-user budget or override it for specific groups, with email alerts as users approach the cap. For AI Credits, the budget controls total usage, not just overage.
- Copilot Max arrives, sign-ups stay paused. A new Copilot Max tier targets power users with higher included usage and limits, while new sign-ups for Student, Pro, Pro Plus, and Max remain paused for now.
What it means for operators
Flat per-seat pricing made AI coding spend easy to forecast and easy to ignore. Usage-based billing ends that. A developer running long agentic tasks or frequent automated code reviews can now cost several times more than a teammate who mostly uses completions, and that variance shows up on the invoice. The teams that handle this well will treat AI tooling like cloud compute: instrument it, set budgets, and review consumption every month.
The practical steps are to turn on user-level budgets before the first full billing cycle closes, decide which workflows justify token-heavy models versus cheaper ones, and keep completions, which stay free, as the default for routine work. If you are scaling an engineering function and want this controlled from the start, it helps to hire an AI engineer who can wire up budgets and model routing, or to build the surrounding AI automation so usage maps to output rather than habit. Priced right, usage-based billing rewards the teams that are deliberate about where the tokens go.
Frequently Asked Questions
June 1, 2026. All Copilot plans now bill on GitHub AI Credits consumed, replacing the previous premium request model.
One AI Credit equals one cent. Each Copilot plan includes a monthly allotment of credits, and usage is drawn down based on token consumption at each model's API rates.
Inline code completions and next edit suggestions remain unlimited on paid plans. Other features, such as chat and agent-style work, consume AI Credits.
Turn on user-level budgets, which are now generally available for organizations and enterprises. Admins can cap per-user spend, receive email alerts near the limit, and route routine work to cheaper models or to the unlimited completions.