GitHub Copilot Pricing Guide 2026: Free, Pro, Pro+, Max, Business & Enterprise Explained

TL;DR

GitHub Copilot's pricing changed more in 2026 than in any year since launch. On June 1, 2026 it abandoned premium-request counting for usage-based billing with GitHub AI Credits. Then it added two brand-new tiers — Copilot Max ($100/mo) in June and Business ($19/user) in July — bringing the lineup to six plans ranging from $0 to $100+.

Base plan prices did not change: Pro stays $10, Pro+ $39, Business $19/user, Enterprise $39/user. What changed is how usage is measured: token-based credits instead of request counts, plus a new flex allotment that adds variable usage on top of your base credits.

Three takeaways: (1) the credit system rewards matching models to tasks — cheap models for routine work, frontier models only when they matter; (2) Business at $19/user is the standout deal for teams, cheaper than Cursor or Windsurf team plans; (3) code completions stay free and unlimited on every plan.

The 6 Copilot Plans at a Glance (August 2026)

Notice something odd? Business ($19) is cheaper than Pro+ ($39) and Max ($100). That's because business plans pool credits across the whole organization — each seat brings an allowance into a shared pool — while individual plans are single-seat. If you're a solo developer, Pro+ or Max is your ceiling; if you're on a team, Business gets you far more for less.

How GitHub AI Credits Work

Before June 2026, Copilot counted "premium requests." A quick chat question and a multi-hour agent session cost the same — which meant heavy users were subsidized and light users paid too much. GitHub replaced this with a token-based system on June 1, 2026:

• AI Credits are dollar-denominated: 1 credit = $0.01 USD. Usage is calculated from token consumption (input, output, and cached tokens) at each model's published API rate.
• Code completions stay free and unlimited on every plan — they never consume credits.
• Chat, agents, cloud agent, CLI, and Spaces consume credits based on the model you use. GPT-5.6 Sol burns credits far faster than GPT-5.6 Luna.
• Copilot code review consumes GitHub Actions minutes in addition to AI Credits.
• Overages: paid plans can buy extra credits beyond their monthly allowance at the same 1 credit = $0.01 rate.

Base credits + flex allotment (new in 2026)

Paid individual plans now split included usage into two parts:

• Base credits — matched 1:1 with your subscription price ($10 for Pro, $39 for Pro+, $100 for Max). These never change.
• Flex allotment — variable additional usage on top of base. GitHub adjusts this over time as model pricing and efficiency evolve. As of August 2026: Pro gets $5 flex ($15 total), Pro+ gets $31 flex ($70 total), Max gets $100 flex ($200 total).

Your base credits are used first; the flex allotment kicks in automatically at the same rates. You don't manage separate buckets — your dashboard shows what's available and what you've used.

The practical impact: a light chat user stays under their plan allowance forever. A heavy daily cloud-agent user on frontier models can burn 7,000+ credits/month ($70 of model usage) — that's exactly who Copilot Max was built for.

Plan-by-Plan Breakdown

Copilot Free — $0

Includes 2,000 completions per month, limited chat and agent usage, and access to models via auto model selection (Haiku 4.5, GPT-5 mini, and more). No credit card required. It's enough to evaluate Copilot, but the daily limits make real agentic work impractical. Verified students get Copilot Student — full premium features at no cost — and teachers and open-source maintainers get Pro free.

Copilot Pro — $10/mo

The sweet spot for most developers. Everything in Free, plus $15 in monthly AI Credits, unlimited code completion and next-edit suggestions, cloud agent and code review access, and the ability to plug in third-party agents like Claude Code and OpenAI Codex. At $10, it remains the cheapest full-featured coding agent on the market — half the price of Cursor Pro and Windsurf Pro at $20.

Copilot Pro+ — $39/mo

Adds premium models including Claude Opus, audit logs, and 4x+ the included usage of Pro ($70 credits/mo). If you run multiple agent sessions daily or need the best models for complex refactoring, Pro+ is where the value is — but note it costs the same as Enterprise per user, with far less team functionality.

Copilot Max — $100/mo (new in June 2026)

For sustained, high-volume agent workflows: $200 in monthly credits (2.9x+ Pro+ usage), priority access to new models and features, and the "you sleep, it fixes bugs" async agent workflow. Realistically only power users who hit Pro+ limits mid-month need this. If you're spending under ~$70/month in credits, stay on Pro+.

Availability note (Aug 2026): new Max sign-ups were temporarily paused during GitHub's reliability push ahead of the June billing transition — existing Student, Pro, and Pro+ subscribers could still upgrade. Check github.com/features/copilot for current availability before deciding.

Copilot Business — $19/user/mo (new in July 2026)

Per-seat pricing with pooled AI Credits across the organization, admin budget controls (user / cost-center / enterprise level), access control, IP indemnity, and data privacy. July 2026 launched this tier for teams that previously had to choose between individual plans or jumping straight to Enterprise. A 50-seat org pays $950/mo for a ~95,000-credit pool.

Copilot Enterprise — $39/user/mo

Everything in Business plus 2x included usage, priority access to new features, and GitHub Enterprise Cloud integration. For larger organizations that need governance at scale.

Model Catalog by Plan (August 2026)

Which models you can actually use is a big part of the value. Verified against GitHub's official plan comparison:

The pattern: Free tiers get auto-selected cheap models; premium models (Claude Opus 4.8, Fable 5, GPT-5.6 Sol) require Pro+ or Enterprise. If your team needs frontier models for architecture work, that's a Pro+ / Enterprise decision — not a Pro decision.

vs Cursor, Windsurf & Claude Code

Copilot's positioning is clear: cheapest entry point, cheapest team plan, and a Max tier that still undercuts Cursor Ultra by $100. The trade-offs are the credit system (you must watch usage) and less deep editor integration than Cursor or Windsurf — Copilot works everywhere, but it's not "AI-first" in any single editor.

New in August: Meta's Muse Code entered the market at $0.20/M output tokens on its Contributor tier — a different billing model (pure usage-based, no subscription). If your Copilot credits regularly run out and you're cost-sensitive on token-heavy agent runs, Muse Code's contributor tier is worth comparing — with the caveat that it trains on your code.

Which Plan Should You Pick?

Casual user → Copilot Free

2,000 completions/month covers light IDE work. If you're a student, teacher, or open-source maintainer, get the free premium tier instead.

Daily professional developer → Copilot Pro

$10/mo with $15 of credits is the best value in AI coding in 2026. Upgrade only when you consistently run out of credits mid-month.

Agent-heavy power user → Copilot Pro+ or Max

If you run cloud agents daily, start with Pro+ ($39). Jump to Max ($100) only if you're burning $70+ in credits monthly — Pro+ includes 4x Pro's usage, which covers most people.

Teams of 2-100 → Copilot Business

At $19/user with pooled credits and budget controls, it's now the cheapest governed team option — half the price of Cursor or Windsurf team plans.

Enterprises with compliance needs → Copilot Enterprise

Audit logs, SSO, fine-tuned models, and 2x Business usage justify the $39/user premium when governance matters more than price.

Bottom Line

The June-July 2026 pricing overhaul made Copilot simultaneously cheaper at the entry level ($10 Pro with credits) and more expensive at the top ($100 Max). The credit system rewards people who match models to tasks — pick cheap models for routine work, frontier models only when they matter, and you'll rarely exceed your allowance. For teams, the new Business tier at $19/user is the standout deal of the year.

Is Copilot worth it in 2026? For most solo developers, yes — $10/month Pro with $15 of credits remains the cheapest full-featured coding agent available, and code completions are unlimited on every paid plan. The two cases where it's not the right pick: (1) you need frontier models daily for complex architecture work — that pushes you to Pro+ at $39, where Cursor Pro at $20 (all-inclusive premium requests) can be comparable value; or (2) you're on a tight budget for token-heavy agent runs — Meta's Muse Code Contributor tier at $0.20/M output is dramatically cheaper, with the data-privacy tradeoff.

One watch-item: the price war isn't over. With Meta's Muse Code undercutting everything at $0.20/M output and Claude Sonnet 5's introductory pricing expiring August 31, expect GitHub to keep tuning flex allotments through the fall. Check the official pricing page before you commit to an annual plan.

How we wrote this

This guide was verified against GitHub's official sources on August 7, 2026:

• GitHub Blog: usage-based billing announcement (May 1, 2026)
• GitHub Blog: flex allotments & Max plan (June 2026)
• github.com/features/copilot — official plan comparison and pricing page
• GitHub Docs: Copilot plans — feature and model matrix

Pricing is current as of August 7, 2026. Flex allotments are variable by design and can change. We have not independently tested every plan tier; plan contents are based on GitHub's published documentation.