GitHub Copilot officially switches to usage-based AI Credits billing today, June 1, 2026 — ending the unlimited premium requests model that millions of developers relied on. Starting today, every Copilot plan gets a monthly allotment of GitHub AI Credits ($0.01 each), and heavy users who burn through their included credits will need to pay for more. It’s a seismic shift for the most-used AI coding assistant on the planet, and developers are already sounding off.
What Are GitHub Copilot AI Credits?
GitHub Copilot AI Credits are a virtual currency priced at $0.01 each that replace the old Premium Request Units (PRUs). Instead of a fixed number of “premium requests,” your usage is now metered by token consumption — meaning longer prompts, bigger codebases, and more powerful models cost more credits per session.
The key change: previously, Copilot plans came with a set number of premium requests per month at a flat rate. Now, every interaction draws down from your credit pool based on how many input and output tokens were processed, plus the model used. Simple tab completions and Next Edit Suggestions remain free — they don’t consume AI Credits at all.
Which Plans Are Affected — And What’s Included
| Plan | Monthly Price | Included AI Credits | Code Completions |
|---|---|---|---|
| Copilot Free | $0 | Limited (new tier) | ✅ Free |
| Copilot Pro | $10/mo | $10 in credits | ✅ Free |
| Copilot Pro+ | $39/mo | $39 in credits | ✅ Free |
| Copilot Business | $19/user/mo | $19/user in credits | ✅ Free |
| Copilot Enterprise | $39/user/mo | $39/user in credits | ✅ Free |
Budget controls give you a safety net: set your additional-spend limit to $0 and Copilot simply stops when credits run out. You cannot be charged more than your set budget.
How It Compares to the Old System
Under the previous model, Copilot Pro+ ($39/month) included 1,500 premium requests per month. Each request — regardless of complexity — counted as one unit. The new system is token-based, so a simple two-line function costs far less than an agent session analyzing your entire codebase. That’s actually good for light users. Heavy users running agentic coding sessions that process millions of tokens per session may find their $39 credit pool drains faster than expected.
Developer reaction has been mixed. A thread on Visual Studio Magazine quoted one engineer: “You will get less, but pay the same price” — reflecting the concern that complex AI coding sessions, which Copilot now heavily promotes, will chew through credits quickly. GitHub counters that the previous flat-rate model was unsustainable as inference costs for longer context sessions escalated dramatically.
Access and Migration Timeline
Monthly Copilot Pro and Pro+ subscribers are automatically migrated today. Annual plan subscribers remain on their existing premium-request terms until their plan expires. Copilot Business and Enterprise customers get promotional included usage for June, July, and August 2026 to ease the transition.
You can monitor your credit usage in Settings → Billing & Plans → Copilot in your GitHub account. Real-time usage dashboards are available for Business and Enterprise admins via the GitHub organization settings.
What This Means for You
If you use Copilot for standard code completions and occasional chat questions, you’ll likely never notice the change — those interactions are cheap to run and your included credits will be more than enough. If you’re running multi-file agent sessions, using Claude Sonnet or GPT-4o via Copilot’s model picker, or running Copilot Workspace for large refactors, watch your dashboard closely for the first few weeks.
For teams on Business plans, this is a good moment to review who actually uses Copilot actively versus who has a seat but rarely opens it. Under the old system, unused seats were simply wasted. Under AI Credits, a team member who never opens Copilot contributes their credit allotment to the org pool — but only if your billing is configured correctly.
The broader trend here is clear: AI coding assistants are becoming infrastructure, and infrastructure gets metered. GitHub’s move mirrors what AWS, Azure, and Google Cloud did when moving from instance pricing to usage-based models. Developers who were comfortable with a flat monthly bill now need to think about AI token costs the same way they think about cloud spend.
Frequently Asked Questions
Will my GitHub Copilot bill go up on June 1?
Not necessarily. Your base subscription price stays the same. You’ll only pay more if you exceed your included AI Credits allotment and have set a positive additional-spend budget. Set your budget to $0 to avoid any surprise charges.
Do code completions (tab suggestions) use AI Credits?
No. Standard code completions and Next Edit Suggestions are excluded from AI Credits billing and remain included in all plans at no extra cost.
Which Copilot models cost the most credits?
Premium models like Claude Sonnet 4.6 and GPT-4o cost more per token than lighter models. Gemini Flash and GPT-4o mini are cheaper per credit. The exact rates are published in GitHub’s Copilot billing docs.
I’m on an annual Copilot plan — am I migrated today?
No. Annual plan subscribers keep their existing premium-request pricing until their plan renewal date. Only monthly subscribers are migrated on June 1, 2026.
What happens when I run out of AI Credits?
If your additional-spend budget is set to $0, Copilot stops processing AI-powered chat and agent requests for the rest of the month. Code completions still work. You can increase your budget at any time in account settings.
GitHub’s full documentation on the new billing model is at docs.github.com. If you’re evaluating alternatives, see our comparison of Claude vs ChatGPT for coding, our breakdown of multi-agent AI frameworks, and our review of Claude Opus 4.8 — the top-rated AI coding model that many GitHub Copilot AI Credits users are comparing it against.
Last updated: June 2026
