What Is GitHub Copilot?
GitHub Copilot is GitHub’s AI coding assistant, built into VS Code, JetBrains IDEs, and the GitHub.com interface. It’s one of the most widely used AI coding tools by raw adoption, but it underwent the biggest pricing shake-up of any major coding assistant in 2026: on June 1, it moved from flat-rate “unlimited” subscriptions to token-based usage billing, and the developer reaction was immediate and loud.
Key Features
- Free inline completions — code autocomplete and Next Edit Suggestions remain free under all plans, unaffected by the billing change.
- Agent mode and chat — multi-step autonomous coding sessions, now metered by token consumption.
- Multi-model access — supports Claude Opus 4.8 (same-day GA at launch), GPT-5.5, GPT-5.4 nano, and Claude Haiku 4.5, each priced differently per token.
- Five plan tiers — Pro ($10/mo), Pro+ ($39/mo), Business ($19/user), Enterprise ($39/user), and a new Max plan ($100/mo) for heavy agentic users.
- Low-resource language support — specialized completion quality improvements for languages like Rust, Haskell, and OCaml.
What Changed on June 1, 2026
GitHub Copilot replaced its old Premium Request Unit system with token-based AI Credits, where 1 credit equals $0.01 of model usage. The base subscription prices didn’t change — Pro is still $10/month, Pro+ still $39 — but the safety net did. Previously, heavy usage was capped by your flat monthly fee; once you exhausted your request allotment, Copilot just fell back to a lighter model so you could keep working. That ceiling is gone. Now, token consumption during chat, agent mode, and code review is metered directly with no fallback, and developers running large agentic sessions have reported bills jumping from $29 to $750, and in some cases from $50 to $3,000.
Each plan ships with a monthly credit allowance: Pro gets 1,500 credits, Pro+ gets 7,000, Business 1,900 per user, Enterprise 3,900 per user. Unused credits do not roll over — they reset to the plan allowance every month.
Who This Doesn’t Affect
If your Copilot usage is bounded by inline completions and Next Edit Suggestions — meaning you mostly use autocomplete rather than chat or agentic features — the June 1 change is effectively a non-event. Your bill stays the same and your editor experience is unchanged. The cost exposure is specifically in chat, agent mode, and code review sessions against large codebases.
Pricing
Subscription prices are unchanged: Pro $10/month, Pro+ $39/month, Business $19/user/month, Enterprise $39/user/month, and the new Max tier at $100/month for heavy agentic workloads. What you’re actually billed for beyond that depends entirely on which model you use and how many tokens your sessions consume — model pricing varies from the cheapest (Claude Haiku 4.5, around $1 input / $5 output per million tokens) up to the most expensive (Claude Opus, $5 input / $25 output per million tokens).
How It Compares
Against Claude Code, Copilot is generally the cheaper entry point if your usage is light-to-moderate, but Claude Code is purpose-built as a terminal-first agentic coding tool with deeper autonomous task execution — see our full Claude Code pricing breakdown for the comparison. Against Cursor, Copilot’s IDE integration is more universal (VS Code, JetBrains) while Cursor offers a more polished dedicated editor experience. The honest framing from multiple independent reviews: Copilot wins on price for light users, Claude Code wins on agentic depth, and Cursor wins on IDE polish — pick based on how you actually work rather than headline pricing alone.
Is GitHub Copilot Worth It?
If you’re a completions-first developer who rarely uses chat or agent mode, Copilot remains genuinely cheap and the June 2026 billing change won’t touch your bill. If you run heavy agentic sessions against large codebases regularly, model your expected token usage carefully before your next billing cycle — the unlimited-feeling flat fee era is over, and costs can scale far faster than the old subscription price suggested.
Last updated: June 20, 2026.