GitHub Copilot Pricing (2026): Free, Pro $10, Pro+ $39, Max $100, and What an AI Credit Buys

The short answer

As of October 19, 2026, GitHub Copilot costs $0 on Free, $10 a month on Pro, $39 on Pro+ and $100 on the new Max plan. For companies, Business is $19 per seat per month and Enterprise is $39. Since June 2026 every plan is metered in GitHub AI Credits at one cent each, and the three paid individual plans include more credits than they cost: Pro gets $15 of usage for $10, Pro+ $70 for $39, Max $200 for $100. Code completions do not touch credits at all.

Below: every plan with its credit allowance, what uses credits and what doesn't, what happens when you run out, and which plan to buy.

Every GitHub Copilot plan

Two lines on GitHub's pages are worth knowing before you budget. The extra credits on the individual plans are a "flex allotment", and GitHub says flex allotments may change over time, so the base number (1,000, 3,900, 10,000) is the floor and the rest is a current bonus. And Copilot is not available on GitHub Enterprise Server, only on github.com and Enterprise Cloud.

What uses AI credits, and what doesn't

One AI credit is $0.01. These draw on credits, at a rate that depends on the model you pick: Copilot Chat, agent mode, code review, the Copilot cloud agent, Copilot CLI, the Copilot app, Spaces and Spark. These don't: code completions and next edit suggestions, which stay unlimited on every paid plan.

That split is the whole pricing story. If you mostly use Copilot as autocomplete, Pro at $10 is effectively unlimited and the credits are a bonus. If you hand it whole tasks in agent mode or from the CLI, the credits are the product, and the model you choose sets how far they go. GitHub doesn't publish a credits-per-task number, so I won't invent one; the model picker shows the relative cost, and lighter models stretch the same allowance further.

One cost sits outside the credit meter: since June 1, 2026, code review workflows also consume GitHub Actions minutes. If your team turns on Copilot code review for every pull request, check your Actions usage as well as your credits.

What happens when you run out

• Individual plans (Pro, Pro+, Max): set a dollar budget for paid usage and Copilot keeps working at a cent a credit, so $10 buys another 1,000 credits. Or switch to a lighter model.
• Business and Enterprise: the admin sets usage limits and decides whether paid overage is allowed. If it isn't, Copilot pauses until the next billing cycle. Credits are shared across the organization, so one heavy user draws down the pool for everyone.
• Alerts: GitHub sends them at 75%, 90% and 100% of any budget you configure, and your Copilot settings show usage and the reset date.

Copilot Pro vs Pro+ vs Max

Per dollar of usage, the ladder gets better as you climb: Pro returns 1.5x its price in credits, Pro+ about 1.8x, Max 2x. The question is whether you'd use them. A worked example: say your agent-mode and chat work uses about 4,000 credits a month. On Pro that's 1,500 included plus 2,500 at a cent each, $10 + $25 = $35. On Pro+ it's $39 with 3,000 credits to spare, plus the premium models Pro doesn't get. So once your overage on Pro would run past about $29 a month, Pro+ is the cheaper plan, and the same math puts the Pro+ to Max switch at about $61 of overage on Pro+ (roughly 13,100 credits a month in use).

Max is for someone running agents most of the day, with priority access to premium models when demand is high. It's the same $100 as Claude Max 5x and ChatGPT Pro's first tier, which is the price point every vendor now uses for heavy agent users.

Business vs the individual plans

A Business seat is $19 and comes with $19 of credits, no flex bonus. On usage alone, a developer gets more from Pro+ at $39 ($70 of credits) than from two Business seats. Business isn't priced on usage, though. It buys central billing, Copilot policy control for the organization, pooled credits with admin spend limits, and GitHub's statement on its plans page that it doesn't use Business or Enterprise data to train its models. If a company is paying for its developers, Business is the plan. If you're a freelancer paying for your own editor, Pro or Pro+ is better value.

Enterprise at $39 per seat requires GitHub Enterprise Cloud and adds a larger credit pool (3,900 per user) plus the enterprise-grade features, such as organization-wide codebase indexing.

GitHub Copilot vs Cursor and Claude Code on price

At the entry level Copilot is now the cheapest serious option. Cursor's lineup changed on October 12, 2026: its $8 Plus plan runs Grok models only, and the cheapest Cursor plan with Claude and GPT is Premium at $30, with those models billed at API list price on top (Cursor pricing). Claude Pro, which includes Claude Code, is $20 a month or $17 billed annually (Claude Code pricing). Copilot Pro is $10 with a selection of models and $15 of credits; the premium models start at Pro+. OpenAI's equivalent is Codex on ChatGPT Plus at $20 (Codex pricing).

Price isn't the whole decision. Which agent writes better code in your repo matters more than $10 a month, and the head-to-heads are on GitHub Copilot vs Cursor and GitHub Copilot vs Claude Code. The full field is ranked on the best AI coding agents.

Which plan to buy

• Free to see whether Copilot fits your editor. 2,000 completions and 50 chats a month is a trial, not a daily driver.
• Pro ($10) if Copilot is mostly autocomplete with some chat. Completions are unlimited and the 1,500 credits cover light agent use.
• Pro+ ($39) if you use agent mode daily or want the premium models. It's cheaper than Pro once your overage passes about $29 a month.
• Max ($100) only if agents run most of your day and Pro+ runs out every month.
• Business ($19 a seat) for any company paying for its developers, for the admin controls and the no-training terms. Enterprise ($39) if you're on GitHub Enterprise Cloud and want the bigger pool.

The subscription is the small number. What costs founders money is what the agent writes and nobody reviews: authentication that only works on the happy path, a key in the client bundle, database rules that let anyone read everything. Before launch, run the free vibe coding security checklist; what the whole path costs, from a $10 plan to a fixed-price rescue, is on what vibe coding actually costs.

Sources

Read on October 19, 2026: GitHub Docs, "Plans for GitHub Copilot" (docs.github.com/en/copilot/get-started/plans: plan prices, base credits, flex allotments, total monthly credits, Business and Enterprise credits per user, Free and Student eligibility, Enterprise Server availability) and GitHub's Copilot plans page (github.com/features/copilot/plans: credit values per plan, "Flex allotments may change over time", what uses credits, unlimited completions, Free's 2,000 completions and 50 chat requests, budgets and alerts, the Business and Enterprise pause rule, the June 1, 2026 Actions-minutes note for code review, the training statement). Cursor prices from cursor.com/pricing and Claude Pro from claude.com/pricing, read the same day. Prices in USD. Per-dollar ratios and the crossover points are my arithmetic from those published numbers.