💡 TL;DR / Summary - GitHub Copilot AI Credits Billing Empirical Key Takeaways (BLUF)
- AI Credits Switch: On June 1, 2026, GitHub Copilot replaces premium request units (PRUs) with token-based usage billing using GitHub AI Credits (1 credit = $0.01).
- Completions Stay Free: Inline autocomplete and Next Edit Suggestions remain unlimited and completely free, consuming zero credits.
- Agent & Chat Metering: Conversational chat, terminal CLI commands, cloud agent loops, and Actions-based Copilot code reviews all consume from your monthly credit allowance (Pro 1,000, Pro+ 3,900 credits).
“Metered token billing is the only viable path for sustaining complex developer agent loops. By aligning price directly with compute (tokens), vendors can offer more advanced models at the cost of actual usage.” — AI Developer Tooling Report 2026
On June 1, 2026, every GitHub Copilot plan drops premium request units (PRUs) and moves to usage-based billing in GitHub AI Credits, where 1 credit equals $0.01. Plan prices stay the same, and each plan keeps a credit allowance roughly equal to its price, but usage is now metered by tokens. Code completions stay free on every plan; the bill risk is in chat and agent sessions, not autocomplete.
This guide separates what actually changes from the noise, using GitHub’s own documentation and announcement. For how Copilot’s model compares to Cursor, Claude Code, and Codex, see our overview of AI coding tool billing; this spoke focuses on Copilot’s credit system. For the same kind of breakdown on Claude Code, see why Claude Code says you’re out of usage. Numbers reflect GitHub’s published rates as of May 2026 and can change, so your Billing Overview page is the final word for your account.
What Are the Three Core Changes Effective June 1, 2026?
The old system counted premium requests. Each model interaction cost one premium request unit, and a multiplier scaled that cost up for more powerful models. A quick chat question and a multi-minute agent run could cost the same single unit, which is exactly the mismatch GitHub says it is fixing.
The new system counts tokens. Every interaction consumes input tokens (what you send), output tokens (what the model generates), and cached tokens (context the model reuses). GitHub prices those tokens per model and converts the total into AI Credits. The practical effect: cost now tracks how much work you actually ask for, not how many times you press enter.
Three structural shifts come with it. First, GitHub is retiring annual plans. Second, the fallback to cheaper models when you ran out is going away. Third, since April 20, 2026, new sign-ups for Pro, Pro+, and student plans are paused, so new users currently land on Copilot Free until that reopens.
AI Credits, in Plain Numbers
The conversion is fixed and simple: 1 AI Credit = $0.01 USD. A $10 budget covers 1,000 credits. Each plan ships with a monthly credit allowance, and GitHub set that allowance at a 1:1 ratio with the plan price.
| Plan | Monthly price | Included AI Credits | Notes |
|---|---|---|---|
| Free | $0 | Small credit allowance + 2,000 completions/mo | Auto model selection; exact free credit amount not published |
| Pro | $10 | ~$10 (≈1,000 credits) | Unlimited completions |
| Pro+ | $39 | ~$39 (≈3,900 credits) | Highest individual access |
| Business | $19/user | $19 | Seat price unchanged; promo bump Jun 1–Sep 1 |
| Enterprise | $39/user | $39 | Seat price unchanged; promo bump Jun 1–Sep 1 |
Once the monthly allowance runs out, paid plans can set an additional budget in US dollars to keep working, shown back to you in credits. If you set no budget, usage stops until the next cycle. The allowance is the floor you already paid for; the budget is the optional ceiling for overflow.
What Burns Credits — and What Stays Free
Two things stay free. Everything else that calls a model draws down credits.
| Stays free (no credits) | Consumes credits |
|---|---|
| Inline code completions | Copilot Chat |
| Next Edit Suggestions | Copilot CLI |
| Copilot cloud agent / agent mode | |
| Copilot Spaces, Spark | |
| Third-party coding agents |
Completions and Next Edit Suggestions remain unlimited on paid plans and never draw down credits. Everything model-driven beyond that does. Because billing is token-based, the heaviest consumers are long-context chats and agent sessions that read and edit across many files. A one-line syntax question costs very little; an agent refactoring a large module reads and writes thousands of tokens and costs accordingly.
Copilot code review adds a second meter. It now runs on an agentic architecture built on GitHub Actions. From June 1, reviewing a pull request with Copilot counts against your included Actions minutes in addition to AI Credits. Your monthly bill can therefore carry two line items from a single workflow.
Monthly vs Annual: Two Different Paths
Your migration depends on how you pay today.
Monthly Pro or Pro+. No action needed. You migrate to usage-based billing automatically on June 1. Your $10 or $39 becomes a credit allowance of the same value.
Annual Pro or Pro+. Your plan does not auto-renew, and it keeps running on premium requests until it expires. There is a catch: starting June 1, model multipliers increase for annual subscribers who stay on request-based billing, so the same work costs more PRUs than before. At expiry you drop to Copilot Free, or you can convert to a monthly plan early and receive prorated credits for the remaining value. GitHub frames annual plans as being phased out, so the long-term path for everyone is monthly usage-based.
Why Your Bill Could Jump
Base prices holding steady is the headline GitHub leads with. The pushback is that a steady price does not mean steady value. Token-heavy workflows like chat, agent runs, and code review become cost-sensitive in a way a flat request cap never was. Developers in GitHub’s own discussion thread raised the same worry repeatedly: the included value drops and the spend gets harder to predict.
How much it moves depends entirely on how you work. Light users who lean on cheaper models for quick chat may find a $10 allowance comfortable, with credits to spare versus the old 300-request cap. Heavy agent users on premium reasoning models are the exposed group. One Pro+ subscriber posted a projection for a plan that cost about €40 a month for 1,500 premium requests. Under the new token rates, their estimate ran far higher. Treat that as a single worst-case self-report, not a typical outcome. The direction is still real: the more agentic and premium-model your workflow, the more the meter runs. Third-party estimates put a typical heavy developer in the $20–$40 per month range once reasoning models are in regular use. That is an outside extrapolation, not a GitHub figure.
The removed fallback matters here too. Under PRUs, exhausting your allowance sometimes dropped you to a cheaper model so you could keep going. Now, when credits and any budget are gone, the work simply stops until reset.
How to Estimate and Cap Your Spend Before June 1
GitHub shipped a preview experience in early May precisely so this is not a surprise.
Open your Billing Overview page on github.com and use the preview tools on the premium request analytics page. “Preview your usage” shows your options, and a downloadable CSV usage report adds two columns that estimate your equivalent cost under usage-based billing next to your current numbers. That report is the most honest answer to “will I pay more,” because it runs on your actual history.
Three levers control spend after migration. Set an overflow budget so usage either caps or continues on your terms rather than stopping unexpectedly. Choose a cheaper model for routine work, since lighter models cost far fewer tokens than premium reasoning models for the same task. And lean on the free lane — completions and Next Edit Suggestions cost nothing, so the more value you get from inline assistance versus long agent runs, the slower your credits drain.
Plan & Credit Reference
| Item | Detail |
|---|---|
| Billing unit | GitHub AI Credits (1 credit = $0.01 USD) |
| Basis | Tokens: input + output + cached, at per-model API rates |
| Free on all plans | Inline completions, Next Edit Suggestions |
| Consumes credits | Chat, CLI, cloud agent, Spaces, Spark, third-party agents |
| Reset | Monthly credit allowance |
| Overflow | Optional USD budget on paid plans; otherwise usage stops |
| Annual plans | Retiring; keep PRUs until expiry, multipliers rise Jun 1 |
| New sign-ups | Pro/Pro+/student paused since Apr 20, 2026 |
Values reflect GitHub documentation as of May 2026 and may change. Your Billing Overview page is authoritative for your account.
FAQ
Q. Will the June 1 change make me pay more?
It depends on your models and how agentically you work. Light chat on cheaper models can fit the included allowance comfortably. Heavy agent use on premium reasoning models is where bills rise, because token-based billing tracks the real compute of long sessions.
Q. Do code completions cost credits?
No. Inline completions and Next Edit Suggestions stay free on every plan, including Free. They never draw down AI Credits.
Q. What is 1 AI Credit worth?
1 AI Credit equals $0.01 USD. A $10 monthly allowance is about 1,000 credits.
Q. I’m on a monthly plan. Do I need to do anything?
No. Monthly Pro and Pro+ migrate to usage-based billing automatically on June 1, with a credit allowance equal to your plan price.
Q. I’m on an annual plan. What happens?
It keeps running on premium requests until it expires, but model multipliers rise on June 1 if you stay on request-based billing. At expiry you move to Free, or you can switch to monthly early for prorated credits.
Q. What happens when my credits run out?
If you set an overflow budget, usage continues against it. If not, usage stops until your next monthly cycle. The old fallback to cheaper models is gone.
Q. Does Copilot code review cost extra now?
Yes, in two ways. It consumes AI Credits, and from June 1 it also counts against your included GitHub Actions minutes, since review runs on an agentic Actions workflow.
Q. Can I still subscribe to Pro right now?
New Pro, Pro+, and student sign-ups have been paused since April 20, 2026, so new users currently get Copilot Free until GitHub reopens them.
Sources
- GitHub Blog — Copilot is moving to usage-based billing: https://github.blog/news-insights/company-news/github-copilot-is-moving-to-usage-based-billing/
- GitHub Docs — Usage-based billing for individuals: https://docs.github.com/en/copilot/concepts/billing/usage-based-billing-for-individuals
- GitHub Docs — Models and pricing for GitHub Copilot: https://docs.github.com/en/copilot/reference/copilot-billing/models-and-pricing
- GitHub Docs — Model multipliers for annual plans: https://docs.github.com/en/copilot/reference/copilot-billing/model-multipliers-for-annual-plans
- GitHub Docs — Prepare for your move to usage-based billing: https://docs.github.com/en/copilot/how-tos/manage-and-track-spending/prepare-for-your-move-to-usage-based-billing
- GitHub Community Discussion #192948 (announcement + FAQ + developer feedback)
Updates & Changelog
- 2026-05-22 — Initial publication. Transition details (PRUs → AI Credits, 1 credit = $0.01, token-based, plan allowances at 1:1 price ratio, free completions, annual-plan path, paused new sign-ups) sourced from GitHub’s blog and documentation. Heavy-bill figures from a user projection and third-party estimates are marked as non-official.
Educational use only. Pricing and plan details change; verify current values on your GitHub Billing Overview page before deciding.
📊Key Empirical Statistics & Metrics
📚Authoritative References & Primary Sources
- GitHub Copilot AI Credits Transition Policy Announcement [External Source]
- GitHub Docs — Usage-based billing for individuals [External Source]