What it is
GitHub’s AI coding assistant, available as a paid subscription with premium-request currency across GPT, Claude, and Gemini models.
Tiers
- Free: 50 premium requests/mo. Enough to test, not heavy use.
- Pro ($10/mo): 300 premium requests/mo. Individual developers.
- Business ($19/user/mo): 300 premium requests/user. Teams under 50.
- Enterprise ($39/user/mo): Requires GitHub Enterprise Cloud at $21/user on top — real cost is $60/user/mo.
Pricing model
Premium-request currency powers Chat, Agent mode, code review, model selection. Heavy agent sessions burn through quota fast — the $10 Pro tier is realistic for casual use, not full-day agent runs.
Verdict
Pro at $10/mo is the lowest-cost entry to multi-model IDE coding. The premium-request abstraction means heavy agent sessions burn through quota fast. Claude Code at $20/mo for similar Sonnet access is arguably better value for serious coding.
See Claude Code vs GitHub Copilot for a detailed comparison.