Updated 29 Sep 2026 • 3 mins read
GitHub Copilot Pricing (2026)
GitHub Copilot pricing in 2026 spans Free, Pro, Pro+, Max, Business, and Enterprise plans, and in June 2026 it switched from premium requests to usage-based AI Credits. This guide breaks down every plan, what draws credits versus what stays free, and how to keep Copilot spend under control.
On June 1, 2026, GitHub quietly changed what you are buying when you pay for Copilot. The plan prices stayed the same, but the meter behind them did not. Out went Premium Request Units, the old count of how many AI actions you got. In came GitHub AI Credits, billed by the token at each model's rate. If you mostly use autocomplete, nothing changed. If you lean on agent mode, chat, and code review, your bill now moves with how much work you hand the model, not how many times you ask.
This guide covers GitHub Copilot pricing in 2026: every plan and price, the new credits model, what still comes free, and how to keep an agentic Copilot from running up a surprise. Prices can change, so confirm current numbers on GitHub's plans page before you buy.
The short version Copilot's individual plans are Free ($0), Pro ($10/month), Pro+ ($39/month), and Max ($100/month); organization plans are Business ($19/seat) and Enterprise ($39/seat). Since June 1, 2026, all plans bill premium AI usage through GitHub AI Credits (1 credit = $0.01), metered by tokens at each model's rate, and each paid plan bundles a monthly credit allowance roughly equal to its price. Code completions and next-edit suggestions stay free and unlimited on every paid plan; agent mode, chat, and code review draw down credits.
The 2026 Change: Requests Out, AI Credits In
The headline shift is billing. Copilot used to meter premium features by counting requests; since June 1, 2026 it meters them by token usage through AI Credits, where one credit equals one cent. A single request might cost one credit or a couple of thousand, depending on the model and how much context it reads, so pricier models spend credits faster. The relief for buyers is that core autocomplete, code completions and next-edit suggestions, stays free and unmetered on every paid plan; only premium model calls, chat, and agent runs draw credits. This is the same move to token-metered billing reshaping the whole category, which we explore in AI costs are cloud costs now.
Individual Plans
Copilot's individual tiers run from free to Max, each bundling a larger monthly credit allowance for premium use.
| Plan | Price | Included AI Credits | Best for |
|---|---|---|---|
| Free | $0 | Limited: about 2,000 completions and 50 premium interactions a month | Trying Copilot; light personal use |
| Pro | $10 / month | $10 a month (GitHub has at times added small flex allowances) | Most individual developers |
| Pro+ | $39 / month | $39 a month, plus premium model access | Heavy chat and agent users |
| Max | $100 / month | About $200 a month (as reported; sign-ups have been paused at times) | Running Copilot as an all-day agent |
official Copilot plans page, since GitHub adjusts allowances and has paused new sign-ups on some tiers during 2026.
Pro at $10 suits most individual developers who want unlimited autocomplete plus some agent and chat use. Pro+ adds headroom and premium models for heavier users. Max only makes sense if you run Copilot as an automated coding agent for hours a day; for most people, Pro or Pro+ is the better value.
Organization Plans: Business and Enterprise
Team plans add pooled credits, admin controls, and, for Enterprise, SSO and codebase customization.
| Plan | Price / seat | Notes |
|---|---|---|
| Business | $19/month | Org-pooled AI Credits, admin and budget controls |
| Enterprise | $39/month | SSO, codebase customization; requires GitHub Enterprise Cloud |
What a Credit Actually Buys: Model Rates Inside Copilot
Since June 1, 2026, a credit is one cent spent on tokens, and Copilot bills every premium interaction at the published rate of whichever model handled it. That makes model choice the biggest lever on how far an allowance stretches. GitHub lists the current rates on its models and pricing reference; the table below shows representative rates as of September 2026, per million tokens, to give a sense of scale.
| Model available in Copilot | Input / 1M tokens | Output / 1M tokens | What it means for credits |
|---|---|---|---|
| GPT-6 Luna | $0.10 | $0.50 | Cheapest option; routine chat and edits barely dent the allowance |
| Gemini 3.8 Flash (promo to Dec 31) | $0.75 | $3.75 | Low-cost workhorse on promotional pricing |
| GPT-6 Sol | $2.00 | $10.00 | Mid-tier default for complex coding |
| Claude Sonnet 5 | $2.00 | $10.00 | Same rate as GPT-6 Sol |
| GPT-6 Astra | $10.00 | $50.00 | Frontier tier; a long agent session can consume a Pro allowance in a day |
A worked example makes it concrete. A typical agent turn that reads 20,000 tokens of context and writes 2,000 tokens costs about 0.3 credits on GPT-6 Luna, about 6 credits on GPT-6 Sol, and about 30 credits on GPT-6 Astra. A $10 Pro allowance is therefore roughly 3,000 Luna turns, 165 Sol turns, or 33 Astra turns. The plan you buy matters less than the model you leave selected.
What Teams Actually Pay: Three Usage Profiles
Because premium usage is metered, two developers on the same plan can pay very different amounts. These three profiles reflect what teams commonly report since the June change, at monthly Pro pricing for an individual and Business pricing per seat.
| Profile | Typical behavior | Individual (Pro $10) | Team seat (Business $19) |
|---|---|---|---|
| Autocomplete-first | Completions all day, occasional chat | About $10; the allowance is rarely touched | About $19; credits pooled and mostly unused |
| Balanced | Daily chat, some agent tasks, a mid-tier model | $10 to $20; allowance roughly consumed | $19 to $35 with modest overage |
| Agent-heavy | Long agent sessions, frontier models, code review on every PR | $40 to $100 or more; Pro+ or Max is cheaper | $60 to $150 or more per seat; budget caps essential |
The third row is where the June change bites. Developers comparing bills have reported large multiples under credit billing for agent-heavy work, and code review on every pull request now runs two meters at once, credits and GitHub Actions minutes. If your team fits that profile, the sticker price is the smallest part of the bill, and the plan decision should be made on modeled usage rather than list price.
What Draws Credits, and What Stays Free
- Free and unmetered on paid plans: code completions and next-edit (autocomplete) suggestions, the core of day-to-day Copilot.
- Draws credits: agent mode, Copilot chat, code review, and any premium model call, billed by tokens at that model's rate.
- Spends faster: more capable models and larger context windows consume more credits per interaction, so routing routine work to cheaper models preserves the allowance.
- After the allowance: premium usage bills at per-credit rates rather than silently downgrading, which is why budget caps matter.
Runs two meters: since June 1, 2026, Copilot code review also consumes GitHub Actions minutes, billed at standard Actions per-minute rates, in addition to AI Credits. Review-heavy teams pay for both, and the Actions charge appears on a different line of the bill.
How to Control Copilot Spend
- Lean on free autocomplete for routine coding, and reserve agent mode and chat for where they clearly save time.
- Route routine premium work to cheaper models to stretch the credit allowance.
- Set per-org or per-seat budget caps so one heavy agent user cannot run up the pooled balance.
- Right-size plans to real usage: many developers do well on Pro, with Pro+ or Max reserved for genuine power users.
The wider discipline of metering and attributing token-based tool spend is the subject of our token economics and TokenOps and LLM cost optimization guides. For another coding agent's economics, see our OpenAI Codex cost and savings guide.Once Copilot is one of several metered AI tools a company pays for, the harder problem is attributing all of that spend to the teams using it, which is what Opslyft's cost visibility is built to do across AI tools and cloud alike.
Copilot vs Cursor vs Codex vs Claude Code: Cost Compared
Copilot's $10 entry point is the lowest of the major AI coding tools, but since all four now meter premium usage, the sticker is only half the comparison. Prices are US list rates as of September 2026.
| Tool | Entry price | Higher tiers | Billing model | Best for |
|---|---|---|---|---|
| GitHub Copilot | Pro $10 / month | Pro+ $39, Max $100; Business $19, Enterprise $39 per seat | AI Credits by token since June 2026; completions unlimited | Teams living in GitHub and VS Code or JetBrains |
| Cursor | Pro $20 / month | Pro+ $60, Ultra $200; Teams $40 or $120 per seat | Credit pool billed at the routed model's list price | Editor-first developers wanting deep agent integration |
| OpenAI Codex | Included in ChatGPT Plus $20 | ChatGPT Pro $100 or $200; Business per seat | Token-based credits on a five-hour window plus weekly cap | Teams already paying for ChatGPT |
| Claude Code | Included in Claude Pro $20 | Claude Max $100 or $200 | Usage limits on a five-hour window plus weekly cap | Complex refactors and long agentic coding sessions |
The Takeaway
Copilot in 2026 is two products wearing one subscription. One is the free, unlimited autocomplete that most developers live in and never think about the cost of. The other is the metered, agentic Copilot, chat, code review, and autonomous tasks, that now runs on a token meter like any other AI service. The plan you pick mostly decides how much of that metered half you get before the credits run out.
So budget around the meter, not the sticker. Know that completions are free, that agent runs and premium models are what cost you, and that a credit is just a cent spent on tokens. Pick the smallest plan whose credit allowance covers your real premium use, cap the pooled balance so no single agent session surprises the team, and Copilot stays the quiet productivity boost it is meant to be rather than a line item that grew a meter overnight.
FAQs
The Copilot Enterprise seat is $39 per user per month, but it requires GitHub Enterprise Cloud at about $21 per user, so the real cost lands closer to $60 per user per month before any premium usage above the included $39 of AI Credits. The promotional extra credits GitHub granted Business and Enterprise customers for June through August 2026 have ended.
On June 1, 2026, Copilot replaced Premium Request Units with usage-based GitHub AI Credits (1 credit = $0.01), billed by tokens at each model's rate. Plan prices stayed the same, but premium usage is now metered by consumption rather than a fixed request count. Code completions remain free.
AI Credits are Copilot's usage currency, where one credit equals one cent, spent by token usage across premium model calls, chat, and agent runs at each model's published rate. Each paid plan includes a monthly allowance, and pricier models spend credits faster.
Yes. Code completions and next-edit suggestions remain free and unlimited on every paid plan and are not metered by credits. Only premium features, agent mode, chat, and code review, draw down your AI Credit allowance.
The Copilot Enterprise seat is $39 per user per month, but it requires GitHub Enterprise Cloud at around $21 per user, so the real cost lands closer to $60 per user per month. Through August 2026, Enterprise seats also receive roughly double the standard AI Credits.
Pro ($10) fits most individual developers who want unlimited autocomplete plus some agent use. Pro+ ($39) adds headroom and premium models for heavier users, and Max ($100) suits running Copilot as an all-day agent. Teams use Business ($19/seat) or Enterprise ($39/seat).
Lean on free autocomplete, reserve agent mode and chat for high-value work, route routine premium work to cheaper models, set per-org budget caps so a heavy agent user cannot drain the pooled credits, and right-size plans to actual usage.
Yes. Verified students, teachers, and maintainers of popular open-source projects get free access to Copilot Pro features through the GitHub Student Developer Pack and GitHub's education programs, which offer more than the standard Free tier. Verification is required, and GitHub has paused new student sign-ups at times during 2026, so check eligibility on the plans page.
On paid plans, premium usage beyond the included allowance is billed as additional AI Credits at each model's per-token rate, at one cent per credit, rather than cutting you off, unless an admin has set a budget cap. Code completions remain unlimited regardless. Subscriptions bought through the GitHub mobile app cannot purchase additional credits.
On sticker price, yes: Copilot Pro is $10 a month against Cursor Pro at $20, and Copilot Business at $19 per seat against Cursor Teams at $40. Both now meter premium usage, so for agent-heavy developers the variable cost can exceed the subscription on either tool. Copilot is the better value for autocomplete-first work inside GitHub; Cursor is often preferred for heavy agentic coding.