GitHub Copilot Moves to Usage-Based Billing: What AI Credits Mean for Your Team

From June 2026, GitHub Copilot is no longer an all-you-can-eat service. The shift to usage-based billing — called “GitHub AI Credits” — is the most significant pricing change since Copilot launched, and it reflects a fundamental truth about AI coding tools in 2026: agentic usage has become the default.

What Changed

Previously, a $10/month Pro plan gave you unlimited “premium request units” (PRUs), whether you asked a quick autocomplete question or ran a multi-hour autonomous coding session. That model was no longer sustainable. With agentic coding agents running for hours in the background, flat-rate pricing meant heavy users cost far more than light users — and there was no mechanism for organisations to control spend.

The new model introduces AI Credits, consumed based on token usage (input tokens, output tokens, and cached tokens) at published API rates. The base plan prices remain unchanged:

• Pro: $10/month
• Pro+: $39/month
• Business: $19/user/month
• Enterprise: $39/user/month

What stays free: code completions and Next Edit Suggestions (NES) — the inline autocomplete features — do not consume credits. What now costs: chat interactions, agentic sessions, code review (which also consumes GitHub Actions minutes), and Copilot Workspace usage.

Practical Implications for Teams

The biggest change is cost predictability. Administrators now have budget controls at enterprise, cost centre, and user level, with pooled included usage across the organisation. This eliminates stranded capacity — unused credits from one team can be consumed by another.

For individual developers, the key behavioural change is awareness. A quick inline completion is free. A thirty-minute agentic refactoring session with multiple tool calls consumes credits. GitHub has discontinued the “fallback experience” (lower-cost model when PRUs were exhausted), so teams that exceed their credit pool see service degradation rather than free fallback.

Business and Enterprise customers received promotional credits for the June-August transition period. Following that, accurate budgeting depends on understanding your token usage patterns.

If you are managing inference costs across your toolchain, the guide to budgeting for LLM inference covers token accounting and caching strategies that apply directly to understanding Copilot credit consumption.