GitHub Copilot Usage-Based Pricing Explained

The short version: if you use Copilot mostly for code completions, your bill probably will not change. If you lean on chat, code review, or coding agents, your subscription price is no longer your ceiling — check your AI Credit usage and set a budget before assuming otherwise. Everything below reflects GitHub’s official documentation as of 2026-07-20.

What changed on June 1, 2026

On June 1, 2026, GitHub replaced premium request units (PRUs) with GitHub AI Credits across all Copilot plans, per the official changelog. Base subscription prices did not change. What changed is the meter: instead of counting requests, Copilot now counts tokens.

Each interaction consumes input tokens, output tokens, and cached tokens. GitHub prices those tokens at the per-model rates in its models and pricing reference, then converts the dollar total into credits at a fixed rate of 1 AI Credit = $0.01 USD. For example, Claude Sonnet 5 is listed at $2.00 per million input tokens and $10.00 per million output tokens, so a short chat on a lightweight model costs far less than a long agent run across many files.

Two things stayed simple. Code completions and next edit suggestions are not billed in AI Credits and remain unlimited on all paid plans. And the credit-to-dollar rate is fixed, so your billing page reads directly in cents.

One special case: since June 1, Copilot code review consumes GitHub Actions minutes in addition to AI Credits. On private repositories, that second meter can matter.

Also new that day: a Copilot Max tier launched for existing individual subscribers, and new sign-ups for Copilot Student, Pro, Pro+, and Max were paused, per the changelog.

Included credits by plan

Official allowances and prices as of 2026-07-20, from GitHub’s billing docs for individuals and organizations:

Plan	Price	Monthly included credits
Free	$0	Allowance included (no published number); auto model selection only; 2,000 completions/month
Pro	$10/month	1,500 (1,000 base + 500 flex)
Pro+	$39/month	7,000 (3,900 base + 3,100 flex)
Max	$100/month	20,000 (10,000 base + 10,000 flex)
Business	$19/user/month	1,900 per seat, pooled
Enterprise	$39/user/month	3,900 per seat, pooled

Base credits match the subscription price and are stable; flex allotments are variable and GitHub says they may change over time.

Pooling works at the billing-entity level: 100 Business seats share a single pool of 190,000 credits, so heavy users draw from what lighter users leave unused.

Promotional allowance: existing Business and Enterprise customers automatically receive a higher temporary allowance from June 1 through September 1, 2026 — 3,000 credits per Business seat and 7,000 per Enterprise seat, pooled. Standard allowances resume afterward.

Credits do not roll over. Unused credits are forfeited, and the allowance resets at 00:00:00 UTC on the first day of each calendar month. Spending beyond the included pool happens only if you (or your admin) enable an additional-usage budget in dollars.

A simple cost example

Say a monthly Pro subscriber consumes 1,850 AI Credits in July. The plan includes 1,500, leaving 350 credits of overage. At $0.01 per credit, that is $3.50, for a pre-tax month of $13.50 ($10 subscription + $3.50 usage). This assumes an additional-usage budget is set; without one, credit-consuming features stop until the next monthly reset.

What happens to annual subscribers

Per GitHub’s announcement, annual Pro and Pro+ subscribers stay on legacy premium-request billing until their term expires — though their model multipliers increased on June 1, 2026. At expiration, the account transitions to Copilot Free unless the user picks a paid monthly plan. Converting early is allowed, and GitHub provides prorated credits for the remaining value of the annual plan.

Who is most likely to notice the new meter

If you mostly use code completions and next edit suggestions on a paid plan, the central AI Credit change is less likely to alter your bill because those two features remain unlimited and outside the credit meter. You should still distinguish them from chat, code review, and coding agents when you read the usage page.

If you regularly use chat or agents, the subscription price is only the starting point. Long interactions can consume more input, output, and cached tokens than short ones, and model rates differ. Your practical unit of comparison is therefore the credit cost of the work you repeat, not the number of prompts you send.

If you use code review on private repositories, include both meters in the decision. Since June 1, the feature consumes GitHub Actions minutes as well as AI Credits. Watching only the credit balance leaves part of that usage out of view.

Organization and enterprise administrators have the broadest budgeting job. Included credits are pooled at the billing-entity level, so a light user’s unused share can be consumed by a heavy user. The temporary Business and Enterprise allowances also end on the stated schedule, after which the standard pooled amounts resume.

Check your account in this order

1. Separate unmetered and metered features. Put completions and next edit suggestions in one group, and chat, code review, and agents in the other.
2. Open the billing overview and identify the expensive features and models. Token-based pricing means two interactions can consume different credit amounts.
3. Compare usage with the included allowance for your actual plan. For an organization, use the shared pool rather than multiplying a personal assumption across seats.
4. Check whether additional usage is enabled. Overage occurs only when you or an administrator has enabled a dollar budget. Without it, credit-consuming features stop at the included limit until reset.
5. Account for the calendar boundary. Credits do not roll over, and the allowance resets at 00:00:00 UTC on the first day of the month.
6. If code review touches private repositories, inspect Actions minutes too. AI Credits alone do not describe that feature’s full meter.
7. If you have an annual Pro or Pro+ plan, note the term end. The legacy premium-request rules continue until then, and the account moves to Free at expiration unless you choose a paid monthly plan.

How to choose between allowance and overage

Stay within the included pool when your credit-consuming work is optional or can pause at reset. In that case, leaving additional usage disabled turns the allowance into the effective ceiling.

Enable a limited dollar budget when continuity matters more than stopping exactly at the included amount. The fixed conversion of one credit to $0.01 makes the overage readable in cents, but the budget should still reflect the amount you are prepared to add to the subscription.

For organizations, do not judge plan fit from the average seat alone. Pooling is useful precisely because use is uneven, but it also means a few heavy users can change the shared result. Review the billing-entity total and the features driving it before changing every seat.

The temporary Business and Enterprise promotion deserves its own scenario. A workload that fits inside the higher allowance through September 1 may not fit after standard allowances resume. Model rates and flex allotments can also change, so the correct plan choice is the one that survives a check against the current documentation, not merely the July snapshot.

The practical move for everyone: open your billing overview, see which models and features consume the most credits, and set a dollar budget you are comfortable with. Flex allotments, token rates, and the enterprise promotion are all date-sensitive, so verify against the official pages before changing plans.