GitHub Copilot pricing: every plan, AI credits, and what a request costs

GitHub Copilot has a free plan, a free Student plan, and three paid plans for individuals: Pro at $10 a month, Pro+ at $39 and Max at $100. Business costs $19 a user a month and Enterprise $39. Since June 1, 2026 every plan comes with a monthly allowance of AI credits, where one credit is $0.01, and chat, agents and code review spend them by the token at each model's rate: Pro includes 1,500 credits, Pro+ 7,000 and Max 20,000. Code completions stay unlimited on paid plans, and Claude Opus 5.5 and GPT-6 Astra need Pro+ or higher. There's no annual billing and no automatic fallback to a cheaper model when credits run out.

Key takeaways

• Plans: Free; Student (free for verified students); Pro $10; Pro+ $39; Max $100; Business $19 a seat; Enterprise $39 a seat.
• Credits: Pro 1,500, Pro+ 7,000, Max 20,000 a month; Business 1,900 and Enterprise 3,900 per user, pooled across the organization.
• Per token: Claude Opus 5.5 is $4 per million input tokens and $20 output, Sonnet 5.5 $2 and $10, GPT-6 Luna $0.10 and $0.50.
• Past the allowance: $0.01 a credit, up to a budget you set; nothing falls back to a cheaper model.
• Unlimited: code completions and next edit suggestions on every paid plan.

How much does GitHub Copilot cost?

From GitHub's plan docs (GitHub):

• Free: 2,000 code completions a month and limited chat and agent use, with automatic model choice. No cloud agent.
• Student, free: for verified students; unlimited completions, the cloud agent and code review, with automatic model choice.
• Pro, $10 a month: unlimited completions, the cloud agent, code review, the CLI and Spaces. Models include Claude Sonnet 5.5, Haiku 4.5, GPT-6 Luna, GPT-5.6 Terra, Gemini 3.8 Flash and Grok, but not Opus, Fable or GPT-6 Astra.
• Pro+, $39: every model, plus third-party agents such as Claude and Codex inside GitHub.
• Max, $100: everything in Pro+, with priority access to premium models.
• Business, $19 a user: organization policies, content exclusion, audit logs, IP indemnity, and no training on your code.
• Enterprise, $39 a user: more credits per user, early access to new models, and Spark; it needs GitHub Enterprise Cloud, which costs $21 a user on top.

Verified teachers and maintainers of popular open-source projects still get Pro free. Prices are monthly only: the $100-a-year Pro plan is gone, and annual subscribers move over at renewal.

How do GitHub Copilot AI credits work?

Each plan's allowance has two parts (GitHub):

• Base credits equal the price and "never change": 1,000 on Pro, 3,900 on Pro+, 10,000 on Max.
• A flex allotment on top, which GitHub says may change: 500, 3,100 and 10,000.

Base credits are used first. Credits reset on the 1st of the month and don't roll over. Every chat message, agent step and review is metered by the tokens it uses at the model's rate, then converted to credits; paid plans get 10% off when they let Copilot pick the model.

When credits run out, extra use costs $0.01 a credit up to the budget you set. "There is no automatic fallback to lower-cost models when a budget is exhausted," so without a budget you stop until the 1st. For individuals GitHub may cap extra usage, and accounts that subscribed through the iOS or Android app can't buy more.

What does each model cost in Copilot?

Per million input and output tokens (GitHub):

• Claude Opus 5.5: $4 and $20. Sonnet 5.5: $2 and $10. Haiku 4.5: $1 and $5. Fable 5.1: $10 and $50.
• GPT-6 Astra: $10 and $50. GPT-6.1 Sol: $2 and $10. GPT-6 Luna: $0.10 and $0.50.
• Gemini 3.8 Flash: $0.75 and $3.75 (a promotional rate through December 31, 2026). Grok 4.7: $2 and $6.

The Claude rates match Anthropic's own list prices. A long agent session on Opus spends credits far faster than chat on GPT-6 Luna, so the model you pick matters more than the plan.

What changed in June 2026?

GitHub announced on April 27, 2026 that Copilot would move to usage-based billing (GitHub). Before June 1, Pro had 300 "premium requests" a month and Pro+ 1,500, each multiplied by a per-model rate, with extra requests at $0.04 and a fallback to an included model when you ran out. Now everything is priced by the token, the fallback is gone, and flex allotments (announced May 12) top up the allowance. New individual sign-ups paused in the spring and reopened from June 17; Business sign-ups reopened September 3.

Copilot Business vs Enterprise

• Pooled credits: 100 Business seats share 190,000 credits a month. Adding seats grows the pool at once; removing them shrinks it next cycle.
• Paid usage is on by default for organizations; an admin turns off the policy to stop it, and can set budgets per user, cost center or organization (GitHub).
• Both include IP indemnity and keep your code out of training; individual plans' data is used for training unless you opt out.
• Enterprise adds 2,000 more credits per user, early model access and Spark, and requires GitHub Enterprise Cloud.

How much do Copilot code review and the cloud agent cost?

• Code review: GitHub estimates $0.05 to $1 worth of credits a review on Lite effort and $0.25 to $5 on Balanced, the default since September 28, 2026 (GitHub). On private repositories it also uses GitHub Actions minutes.
• Cloud agent: credits for the tokens it uses plus Actions minutes, with sessions capped at 59 minutes. GitHub doesn't publish a typical cost per session.

How Copilot's reviewer compares with Claude's and others is in AI code review tools.

Copilot vs Claude Code vs Cursor

• Copilot Pro, $10: 1,500 credits across many models (no Opus), unlimited completions, the cloud agent and code review.
• Claude Pro, $20 ($17 annually): includes Claude Code with five-hour and weekly limits instead of credits; Opus and Sonnet are included (Claude Code pricing).
• Cursor Pro, $20: a large pool for Cursor's own models plus other models at API prices (Cursor pricing).

The full head-to-head is in Claude Code vs Copilot, and other options in GitHub Copilot alternatives.

One agent a whole team can steer

Copilot's credits and sessions belong to each person. Poly (usepoly.co) is a shared browser room where a team works with one Claude Code or Codex agent: everyone sees each step live, and any member can approve a change before it happens. Free to start. What is Poly?