GitHub Copilot Pricing and AI Credits: What Free, Student, Pro, Pro+, Max and Business Include

GitHub's plan table has a column of numbers labelled "GitHub AI Credits", and it is easy to read that column as a request count. It is not. A credit is a spend allowance, the paid plans split it into a fixed part and a variable part, and the two free plans do not publish a number at all. Work out which unit your account uses before you compare prices or copy a classmate's limit.

What each Copilot plan costs

Plans for GitHub Copilot is the source for this table. GitHub also sells a larger organization plan on different credit and price terms; it is outside the scope of this comparison. Copilot is not available for GitHub Enterprise Server.

Two rows deserve more attention than the rest.

Free and Student do not have a published credit number. GitHub's table says each includes "an allowance of GitHub AI Credits" and stops there. If you have seen a specific figure for the student allowance, it did not come from GitHub's current plan documentation. What GitHub does publish is that Copilot Free includes 2,000 code completions per month, that Copilot Student includes unlimited code completions, and that both plans reach models through auto model selection only, with Free limited in agents and Student excluding third-party agents. Copilot Student is available to verified students, and Free is for individual developers who do not already have Copilot through an organization. Usage-based billing for individuals

The paid plans split the allowance in two. Base credits are included with the subscription every month, match the subscription price, and never change. The flex allotment sits on top and is the variable part, designed to move with model pricing, new models and efficiency gains. Your base credits are consumed first, and the flex allotment then applies automatically, at the same rates across your IDE, on GitHub.com and in Copilot CLI, with no setup step. Copilot AI credits allowance by plan

One credit is one cent, and what spends it

GitHub converts model-priced input, output and cached tokens into AI credits at a fixed rate: one AI credit equals $0.01 USD. The cost of an interaction therefore depends on the model used and the number of tokens consumed. A short question to a lightweight model may cost a fraction of a credit; a long cloud agent session across several files with a frontier model costs more, because it is doing more work.

Three factors move the number faster than most people expect: conversation length and complexity, agentic work that makes several model calls inside one task, and the model you choose. A count of the messages you typed will miss most of an agent session's consumption. Paid users also get a 10% discount on model costs while using auto model selection in chat, the CLI, the app or the cloud agent. What affects my usage

The reset follows the calendar, not your invoice

Included AI credits do not carry over. Unused credits are forfeited, and the allowance returns to the full monthly amount at 00:00:00 UTC on the first day of each calendar month. GitHub states that this reset date is fixed and does not change based on your subscription billing date.

That produces a specific and confusing result. If you subscribe on the 20th, your first billing cycle runs from the 20th to the 19th of the next month, but the credit allowance resets on the 1st. Your first cycle therefore contains a short allowance period and a long one, and the second contains the reverse. Keep billing dates and allowance reset dates separate when reviewing a usage counter. Billing cycles for metered products

Business is a shared pool, not a personal bucket

This is the row most often misread. A Copilot Business license comes with 1,900 AI credits per user per month, and those credits are pooled at the billing entity level. GitHub's own example: an enterprise with 100 Copilot Business users gets a shared pool of 190,000 AI credits rather than 100 individual buckets. Power users draw more when they need it and lighter users offset that consumption. Usage-based billing for organizations and enterprises

Four controls sit on top of the pool, and they are easy to confuse with one another:

• User-level budgets cap what an individual may consume, from the shared pool and from additional usage. A $0 user-level budget blocks that user immediately.
• Cost-center budgets cap metered charges for a defined group of users after the pool is exhausted.
• Enterprise spending limits cap total metered charges across the whole enterprise after the pool is exhausted.
• Organization-level budgets cap metered charges for seats billed to the organization, also after the pool is exhausted.

A user who exhausts their own budget loses access even if the shared pool still has capacity, and there is no automatic fallback to a cheaper model. Additional usage is enabled by default for organizations; an administrator has to disable the "AI credits paid usage" policy to stop it. If additional usage is not allowed, usage is blocked until the next cycle. Budgets are set in US dollars and draw down at the same fixed rate, so a $10 budget covers 1,000 AI credits.

When you run out

One administrative detail catches people out: if you hold a personal Copilot Pro, Pro+ or Max plan and are then assigned a seat in an organization, your personal plan is cancelled automatically and you receive a prorated refund for the rest of the cycle. About changes to your Copilot plan

If your usage numbers look wrong

GitHub's own documentation warns that older clients display incorrect model pricing, inaccurate usage information or outdated billing terminology, and that usage alert notifications may not appear as expected. The minimum versions it lists are VS Code 1.120, Visual Studio 2022 17.14.33, Visual Studio 2026 18.6.0, SQL Server Management Studio 22.6, the JetBrains plugin 1.9.1, the Eclipse plugin 0.18.0, the Xcode extension 0.50.0 and Copilot CLI 1.0.48. Before concluding that a plan gives you less than it should, update the client and check the figure again. Update your IDE, client and extension

How this differs from legacy premium requests

Some annual Pro and Pro+ subscriptions still run on the older request-based billing, where usage is counted in premium requests with per-model multipliers rather than in credits. Advice about request multipliers will not explain a credit-based bill, and a credit allowance will not explain a request counter. GitHub Copilot Premium Requests vs AI Credits covers that older system, the transition and the annual-plan exception.

Check one task before buying more capacity

Choose a task with a defined end: adding a test for an existing bug, walking a dependency upgrade, reviewing one pull request. Note the files it touches and the check that proves it worked. Run it on your current plan, and record the credit consumption afterwards, not the number of messages you typed, because an agent session spends most of its credits inside the session. Then run the same task on the plan you are considering.

That comparison tells you something a plan table cannot: whether the constraint is your allowance, your model choice or the shape of the task. If the allowance is what stops you, a higher tier helps and a cheaper model choice inside the same tier also helps. If the task itself is expensive, a bigger allowance only delays the same wall.

Sources

• Plans for GitHub Copilot
• Usage-based billing for individuals
• Usage-based billing for organizations and enterprises
• Budgets for usage-based billing
• Setting up budgets to control spending on metered products
• Optimizing your AI usage to maximize efficiency and reduce cost
• GitHub Copilot Premium Requests vs AI Credits

The plan prices, credit allowances, reset rule and pooling behavior above were read from GitHub's documentation on 2026-10-20. GitHub can change prices, allowances and client requirements, and the flex allotment in particular is designed to move. This article does not describe what any specific account will be granted.