GitHub Copilot budget expiration became generally available on September 1, 2026, giving Business and Enterprise customers a way to end temporary individual spending overrides automatically. The change is relevant to QA teams running short AI coding or test-generation pilots.
What changed
According to GitHub’s announcement, administrators can leave an individual budget without an expiry, end it at the next billing-cycle boundary, or choose a specific date. They can also change or remove the expiry later. The billing interface supports the setting, and API integrations can use the expires_at field when creating or updating budgets.
When the date arrives, GitHub removes the individual budget. The user then inherits the applicable cost-center per-user budget, or otherwise the universal user budget. Previously, temporary overrides required manual removal.
Check the fallback before the pilot ends
GitHub’s budget documentation clarifies an important edge case: if neither fallback is configured, no user-level budget applies after expiry. Other applicable spending controls still need consideration. The documentation also says user-level budgets enforce a hard stop and can block a user even when broader budgets have capacity.
Why this matters for QA engineers
Our practical takeaway: a temporary budget increase for a test-generation experiment can now have a defined end date. For example, a QA lead could align an individual override with the close of a regression-automation pilot, then confirm which normal budget will apply.
Before adopting the setting, record the intended fallback, validate the expiry configuration with your billing administrator, and include budget-blocked runs in pilot reporting. An interrupted assistant session should be distinguished from a failed test or poor model output. These are QA recommendations, not performance claims from GitHub.
Verified September 8, 2026. This report covers the September 1 announcement; it does not describe a new model release.