GitHub Copilot's billing model changed on June 1, 2026. Every plan—Free, Pro, Pro+, Business, Enterprise, and the new Copilot Max tier—now runs on GitHub AI Credits instead of premium request units (PRUs). One AI credit equals $0.01 USD. Usage is calculated from token consumption—input, output, and cached tokens—at each model's published API rate.
What Replaced PRUs
Under the old model, "premium requests" were an abstraction layer on top of model multipliers. The new model is closer to raw API economics: pick a model, consume tokens, convert to credits. Autocomplete and next-edit suggestions remain unlimited on paid plans and do not draw from AI credits—the billing change targets chat, agent, and code-review workloads.
Individual Plan Allowances
Each paid individual plan includes two pools per month:
- Base credits — fixed with your subscription (matches subscription price in dollars).
- Flex allotment — additional included usage GitHub can adjust as model pricing and efficiency evolve.
Base credits spend first; flex applies automatically at the same rates across IDE, GitHub.com, and Copilot CLI. Totals for June 2026:
- Copilot Pro ($10/mo): 1,000 base + 500 flex = 1,500 credits
- Copilot Pro+ ($39/mo): 3,900 base + 3,100 flex = 7,000 credits
- Copilot Max ($100/mo): 10,000 base + 10,000 flex = 20,000 credits — new tier for Student, Pro, and Pro+ upgrades
After included credits exhaust, set an additional spending budget in dollars to keep working; usage draws down at 1 credit = $0.01. GitHub may cap additional spend for individual accounts based on verification and billing history.
Organization and Enterprise Changes
Business ($19/user/mo) and Enterprise ($39/user/mo) licenses include per-seat credits pooled at the billing entity. Through August 2026, existing Business and Enterprise customers receive promotional included usage—$30 and $70 in monthly credits respectively—above the standard allotments.
New controls shipped alongside billing: user-level budgets for orgs and enterprises (GA), org-wide default runners for Copilot code review, and granular budget overrides per user group.
Code Review Bills Twice Now
Copilot code review consumes both AI credits and GitHub Actions minutes. The model inference draws credits from the requester (or PR author on auto-triggered reviews); the agentic infrastructure runs on Actions minutes attributed to the repository. Teams that enabled automatic review on every PR should model both costs—not just token usage.
Migration Notes
- Monthly Pro and Pro+ subscribers migrated automatically on June 1.
- Annual Pro and Pro+ plans stay on request-based pricing until expiry, with higher model multipliers from June 1; they convert to monthly credit billing at renewal.
- New individual plan sign-ups were paused at launch; GitHub indicated reopening in the coming weeks.
- Claude Fable 5 arrived in Copilot's model picker on June 9 but was suspended June 12—model availability and billing math can now change independently.
Why It Matters for Web Developers
If your Copilot budget was set when PRUs masked token economics, recalculate now. Long agent sessions on expensive models burn credits faster than the old multiplier table suggested—GitHub's own Fable 5 benchmarks claimed fewer tool calls and lower token use on autonomous workflows, which directly affects cost under this model.
Three moves this week: open the usage dashboard and confirm base versus flex consumption; set org budgets before enabling Copilot agents team-wide; and compare per-token rates in GitHub's model pricing docs against routing the same work through Claude Code or Cursor if your team runs multi-tool. Copilot's moat is IDE breadth and enterprise procurement—not necessarily the cheapest token path anymore.