GitHub Copilot costs $0, $10, $39 or $100 a month for an individual, and $19 or $39 per user for a team. Those are the numbers on the pricing page. They are not the numbers that decide what you can actually do with it.
What decides that is a unit GitHub introduced quietly and explains in one line of its FAQ: 1 AI credit = $0.01 USD. Once you know that, every plan on the page converts into a dollar figure, and the shape of the pricing becomes obvious — including one thing GitHub does not draw attention to, which is that individuals are sold AI below cost and teams are not.
Every plan, and what it is really worth
| Plan | Price / user / month | Monthly AI credits | Credits in dollars |
|---|---|---|---|
| Free | $0 | An allowance, not published | — |
| Pro | $10 | 1,500 | $15 |
| Pro+ | $39 | 7,000 | $70 |
| Max | $100 | 20,000 | $200 |
| Business | $19 | 1,900 | $19 |
| Enterprise | $39 | 3,900 | $39 |
Read the last column downwards. On Pro you pay $10 and are handed $15 of model usage. On Pro+ you pay $39 and get $70. On Max you pay $100 and get $200 — double. Then the pattern stops dead. Business is $19 for $19. Enterprise is $39 for $39. Exactly face value, to the cent.
This is not a complaint about the team plans. It is a description of what they sell. The money above face value on Business and Enterprise buys pooling, policy controls, audit, IP indemnity and support. It does not buy more model time per dollar. If you are choosing between a Business seat at $19 and a personal Pro at $10 purely on how much Copilot you get, the $10 plan gives you 79% as many credits for just over half the money.
The part of your allowance GitHub can change
The totals above are made of two different things, and only one of them is promised. GitHub’s billing documentation splits them:
| Plan | Price | Base credits | Flex allotment | Total |
|---|---|---|---|---|
| Pro | $10 | 1,000 | 500 | 1,500 |
| Pro+ | $39 | 3,900 | 3,100 | 7,000 |
| Max | $100 | 10,000 | 10,000 | 20,000 |
Base credits, in GitHub’s words, “are included with your plan subscription each month. These match with your subscription price and they never change.” Note what that sentence quietly confirms: your guaranteed allowance is exactly the money you hand over. $10 buys 1,000 credits, which is $10.
Everything above that line is the flex allotment, described as “a variable part of your included usage… designed to adapt as the economics of AI evolve, including model pricing, new models, and improvements in efficiency.” The plans page puts it more bluntly in a footnote: “Flex allotments will vary over time.”
So a third of Pro’s advertised allowance, 44% of Pro+’s and half of Max’s is a figure GitHub has reserved the right to move. That is a reasonable way to run a business whose input costs are volatile. It is also a reason not to build a team’s capacity planning on the headline number. Plan against base credits; treat flex as weather.
What credits actually pay for — and what stays free
This is the most commonly misunderstood part of Copilot pricing, and it works in your favour.
Credits are not spent on autocomplete. GitHub is explicit: “Code completions and next edit suggestions are not billed in AI credits and remain unlimited for all paid plans.” On organisation plans they survive even a hard stop — completions “remain available when a user has a $0 USD budget or reaches an enforced spending limit.”
Credits are spent on the conversational and agentic surface: Copilot Chat, Copilot CLI, the Copilot cloud agent, Copilot Spaces, Spark, and third-party coding agents.
The practical consequence is that a developer who mostly uses Copilot as very good autocomplete may never touch their allowance at all. The developer who hands a multi-file refactor to a cloud agent on a frontier model can spend a meaningful fraction of a month’s credits in one afternoon. The cost of an interaction, per the docs, comes down to the model chosen and the number of input, output and cached tokens consumed.
Three things that will surprise you on the invoice
1. Nothing rolls over, and the reset is not your billing date
“Included AI credits do not carry over between months. Unused credits are forfeited, and your allowance resets to the full monthly amount at 00:00:00 UTC on the first day of each calendar month. This reset date is fixed and does not change based on your subscription billing date.”
Subscribe on the 25th and you get six days of that month’s allowance before it resets. The subscription charge does not prorate to match. It is a small amount of money and a large amount of confusion, so it is worth knowing before you sign up mid-month rather than after.
2. On team plans, overage is on unless somebody turns it off
GitHub states it plainly: “On Copilot Business and Copilot Enterprise, additional usage is enabled by default. If you want to prevent any spending beyond your included AI credits, an administrator must explicitly disable the AI credits paid usage policy.”
If nobody in your organisation has been into the AI Controls settings, your Copilot bill has no ceiling. That is a five-minute job for whoever owns the budget, and it is the single most valuable thing in this article for a team lead.
3. Usage is billed after it happens, including after you cancel
“AI credits usage beyond your plan’s included amount is billed after it occurs, separately from your subscription renewal… Usage from your final days on the plan might not appear on your last invoice, so you might receive a charge after you cancel.”
How the team pool works
Business and Enterprise credits are not per-person buckets. They are pooled at the billing entity, and GitHub’s own example is the clearest statement of it: “an enterprise with 100 Copilot Business users gets a shared pool of 190,000 AI credits rather than 100 individual buckets. This means power users can draw more when they need it, while lighter users offset that consumption.”
For most teams this is strictly better than per-seat allowances, because real usage is never evenly distributed. Two caveats. Adding licences mid-cycle grows the pool immediately, but removing them does not shrink it until the next cycle — so a mid-month reduction does not save you credits, only the next month’s seat fees. And a heavy agent user can quietly consume a disproportionate share of a pool that everyone is sharing, which is what the budget controls are for.
One trap sits outside both plans. Copilot access through GitHub Enterprise Cloud, or as Copilot Standalone, “has no included allowance. Metered usage is billed from the first use.” If your organisation reaches Copilot that way, there is no free pool underneath you at all.
What the free plan is genuinely good for
Copilot Free gives you 2,000 code completions a month, an unpublished allowance of AI credits, and access to models “through auto model selection only” — you cannot choose the model. Named on the plans page are Haiku 4.5 and GPT-5 mini. There is no credit card requirement.
Two thousand completions is a real constraint rather than a token one. A developer working steadily in a well-typed codebase can see a few hundred suggestions in a day. Treat Free as a fortnight’s evaluation rather than a permanent arrangement — which, to be fair, is what it is for.
Verified students do better than everyone: the Student plan includes unlimited code completions, where Free is capped at 2,000. If you are eligible, that is the single biggest free allowance on the page.
Which plan fits which developer
- You mostly want autocomplete. Pro, $10. Completions are unlimited on every paid plan, so the more expensive tiers buy you nothing you will use.
- You live in Copilot Chat and run agents daily. Pro+ at $39 gives 7,000 credits against Pro’s 1,500. If you are regularly topping up a Pro plan by more than $29 a month, the upgrade has already paid for itself.
- You run long agent sessions across large repositories. Max, $100 for 20,000 credits. GitHub positions it for “sustained, high-volume agent workflows” and it is the only individual tier that doubles your money in credits.
- You are buying for a team. Business at $19. You are paying roughly a Pro subscription extra, per seat, for IP indemnity, access and budget control, governance and a shared pool. Those are the right things to buy for a team; just do not expect extra model capacity with them.
- You need org-wide rollout and the largest pool. Enterprise at $39, which roughly doubles the per-seat credit contribution and adds priority access to new models.
Two discounts with dates on them
GitHub’s docs currently carry two reductions in model cost. One is permanent-looking: paid plans “qualify for a 10% discount on model costs while using auto model selection in Copilot Chat, or Copilot cloud agent.” Letting Copilot pick the model is cheaper than picking it yourself.
The other has an expiry: “Through November 30, 2026, users on paid Copilot plans receive a 30% discount on model costs when using Copilot Fusion.” If you are reading this in December, assume that one is gone and check before you budget on it.
How to keep the bill where you want it
- Set a budget, in dollars. Additional usage draws down at $0.01 a credit, so a $10 budget is 1,000 credits. GitHub sends alerts at 75%, 90% and 100%.
- Admins: disable paid usage if you want a hard stop. It is on by default on Business and Enterprise.
- Drop to a lighter model for routine questions. GitHub’s own advice. A quick question to a small model costs a fraction of a credit; the same question to a frontier model does not.
- Upgrade rather than top up, if you are doing it often. When you move tiers you pay only the price difference, and earlier usage in the cycle is counted inside the new, larger allowance.
- Watch code review. It runs on GitHub Actions and consumes Actions minutes, which is a separate line on your bill from AI credits.
- Check who else is spending your credits. A GitHub App installed on your organisation can make Copilot requests using an installation token, and the organisation that owns the installation pays. You cannot set a budget for an individual app.
Frequently asked questions
Is GitHub Copilot free?
Yes, there is a genuinely free plan with no credit card required. It gives you 2,000 code completions a month, an allowance of AI credits that GitHub does not publish, and model access through automatic selection only. Verified students get unlimited completions instead of 2,000.
How much is GitHub Copilot Pro?
$10 per user per month. It includes unlimited code completions and next edit suggestions, model selection, access to third-party agents including Claude Code and Codex, and 1,500 AI credits — worth $15, since one credit is one US cent.
What is a GitHub AI credit?
It is GitHub’s billing unit for model usage, fixed at $0.01. Your interaction’s input, output and cached tokens are priced according to the model you used, and the total is converted into credits. A short question to a lightweight model can cost a fraction of a credit; a long agent session on a frontier model across many files costs far more.
Do unused Copilot credits roll over?
No. Unused credits are forfeited and the allowance resets at 00:00:00 UTC on the first day of each calendar month, regardless of when your subscription bills. The same applies to the shared pool on Business and Enterprise.
What happens when Copilot credits run out?
Code completions keep working — they are not billed in credits. For chat and agents you can upgrade your plan, set a dollar budget to continue at $0.01 per credit, switch to a cheaper model, or wait for the monthly reset. On team plans, whether you can continue at all depends on the policy your administrator has set.
Is Copilot Business worth it over Copilot Pro?
It depends what you are buying. Per dollar, Business gives you less model usage than Pro — $19 of credits for $19, against $15 of credits for $10. What it adds is a shared credit pool, access and budget controls, governance, IP indemnity and data privacy terms, and web-based support. For a company those are usually the deciding factors, and the credits are secondary.
What we checked
GitHub’s Copilot plans and pricing page, its usage-based billing documentation for individuals and the equivalent page for organisations and enterprises were read on 10 November 2026, and everything in quotation marks is taken verbatim from them. Prices are the published US dollar figures and exclude tax. The dollar values of each credit allowance are our own arithmetic from GitHub’s stated rate of $0.01 per credit, shown so you can check them. We have not tested Copilot, and nothing here is a claim about how well it writes code.
For how Copilot compares with the other coding assistants, see Cursor vs GitHub Copilot vs Aider vs Devin Desktop. For credit systems elsewhere, see What AI credits actually buy, tool by tool and When the credits run out, what actually stops? The directory entry is GitHub Copilot.