How to keep Copilot bills predictable after the June 1 switch

Published 2026-05-14 by Owner

GitHub Copilot switches to AI Credit billing on June 1. Plan prices don’t change, but most features beyond code completions will now draw from a monthly Credit allotment. Here is what to know before the switch.

For context on why GitHub made this move, see The Copilot pricing shock.

Which features drain Credits fastest

Not all features cost the same. From highest to lowest Credit consumption:

Cloud agent sessions — autonomous multi-step tasks. Longest runtime, highest compute, highest Credit cost. A 20-minute session on a complex task will cost significantly more than a single Chat exchange.
Code review in GitHub Actions — starting June 1, each code review job consumes Actions minutes plus Credits. Teams with high PR volume will see this add up.
Copilot Chat — per-message cost, varies by model selected. GPT-5.5 and Claude Opus 4.7 cost more than GPT-4o or Sonnet.
Copilot CLI — each command invocation. Lower cost than Chat per interaction, but CLI use tends to be frequent.

Code completions and Next Edit suggestions do not consume Credits. These remain effectively unlimited on all plans.

How to set a spending cap

GitHub lets you set a hard monthly spending cap per plan. When you hit the cap, paid features stop until the month resets — completions still work.

For individual plans (Pro, Pro+):

Go to github.com → Settings → Billing and plans → Spending limits
Set “Copilot” spending limit to $0 to prevent any overage, or to a specific dollar amount to cap it
You will receive an email warning when you reach 75% and 100% of your allotment

For team plans (Business, Enterprise):

Go to Organization Settings → Billing → Spending limits
Set a per-organization monthly cap
Admins receive usage reports; individual engineers do not see other engineers’ Credit usage by default

What to turn off if you’re on the base plan

Copilot Pro ($10/month) includes the smallest Credit allotment. If you’re on the base plan and do significant Chat or agent work, you may exhaust Credits before the month ends.

Turn these off or limit them to extend your allotment:

Code review in Actions — disable the copilot-review workflow trigger in your .github/workflows files if you’re not actively using it
Copilot in CLI — uninstall the extension with gh extension remove github/gh-copilot if you don’t use it regularly
Copilot Spaces — these consume Credits on creation and active use; archive idle spaces

Leave these on — they cost nothing:

Code completions (Tab / Next Edit)
Copilot in the IDE toolbar — the toolbar itself is free; only Chat sessions cost Credits

For teams: getting per-seat visibility

Business and Enterprise admins can see aggregate Credit usage per user in Organization Settings → Copilot → Usage. This view shows which users are consuming the most Credits and from which features.

If one engineer is running extended cloud agent sessions, they’ll appear as a disproportionate share of the team’s Credit consumption. This is normal for engineers doing agentic-heavy work (migrations, refactors, test generation) — the visibility helps distinguish “high usage, high value” from “high usage, unclear value.”

The one thing not to turn off

Code completions stay free and remain the highest-ROI feature in Copilot’s suite. Every benchmark and usage study shows inline completions and Next Edit suggestions as the features with the most consistent daily impact. Keep those on regardless of Credit pressure.