Copilot Seat Prices Held. The Credit Meter Did Not.
GitHub moved Copilot from request-based billing to usage-based billing on 1 June 2026. Seat prices stayed: Copilot Pro at $10 a month, Pro+ at $39, Business at $19 per user per month, Enterprise at $39 per user per month. Paid plans now include a monthly allotment of GitHub AI Credits at $0.01 a credit. Overflow is metered. Completions and Next Edit Suggestions stay unlimited on paid plans. That is GitHub Copilot credits: a seat plus a meter.
That’s the problem.
This is not a list-price event. It is a unit change. Do not let Microsoft frame the next conversation as “your seat price stayed the same.” The seat is still the floor. The bill is now tokens (input, output, and cached) converted into credits at the listed API rate for the model you actually called. Completions still look like a seat. The invoice does not.
What GitHub actually published
On 27 April 2026 GitHub announced that all Copilot plans would move to usage-based billing on 1 June. The changelog confirmed the switch is live.
GitHub’s own rationale is on the blog. Copilot is no longer an in-editor assistant. It runs long, multi-step coding sessions. “Today, a quick chat question and a multi-hour autonomous coding session can cost the user the same amount. GitHub has absorbed much of the escalating inference cost behind that usage, but the current premium request model is no longer sustainable.”
The new unit is GitHub AI Credits, worth $0.01. Customers consume input, output, and cached tokens, converted at the listed API rates for each model. Paid plans can purchase additional usage. Code completions and Next Edit Suggestions do not consume credits. They remain unlimited on paid plans.
Included monthly credits match the seat in dollar terms. Pro: 1,000 credits ($10). Pro+: 3,900 ($39). Business: 1,900 per user ($19). Enterprise: 3,900 per user ($39). Existing Business and Enterprise customers received a higher allotment from 1 June through 1 September 2026: 3,000 and 7,000 respectively. After that date those users drop to the standard pool. Credits do not carry over. The pool resets at 00:00 UTC on the first of each calendar month. Business and Enterprise credits are pooled at the billing entity, not trapped per user.
Once the allotment is gone, you set an overflow budget or usage stops until the cycle resets. Additional usage is enabled by default for organizations. If you want a hard cap, an admin has to turn paid usage off. There is no automatic fallback to a cheaper model.
The docs publish per-million-token rates by model. This post will not invent a dollar-per-prompt rate on top of that table.
GitHub Copilot credits are a seat plus a meter
If that feels familiar, the problem is not your team’s maturity. You have been managing a Copilot seat. GitHub is now selling a seat, a credit pool, and a model mix. The model is the commodity when Claude sits inside Copilot; from Copilot to coworker is the workforce overlay if those credits are a digital employee, not a developer perk. A reporting layer that only prints named users will miss the bill. A FinOps export that only prints tokens will miss the seat floor you still pay, and the unlimited completions that still travel with it.
This is not a criticism of the teams building those reports. Traditional SAM tools collect comprehensively and answer slowly. They were built to prove who has a seat, not to recast that seat against a credit burn you cannot know in advance. The old request was a blunt unit. The new unit tracks inference, and usage-based billing is non-deterministic.
Picture your procurement director in a mid-year Microsoft review. The vendor walks in and says the SKU price did not change. Your CIO wants to know why Copilot is over the line. If that answer takes three weeks of specialist reading across a GitHub admin console, a usage export, and a model-mix spreadsheet, you have already lost the framing of the meeting.
An analysis that arrives after the salesperson has named your unchanged SKU is a document. One that arrives during the meeting is a decision input.
The model mix is the price increase that never hit the SKU
Seat prices did not move. The cost of calling a premium model did.
GitHub’s models and pricing table is the live rate card. Lightweight models are cheap. Frontier Anthropic and OpenAI models are not. A long agent session on Opus or GPT-5.5 will burn the pool in a way a completion never will. Completions stay unlimited. Chat and agents do not.
If your file still has one Copilot cell with a seat count in it, you are looking at a floor, not a license position.
What you should have locked before 1 June (and what is still live)
1 June already landed. The lesson did not expire. Existing Business and Enterprise customers still sit on the grace allotment through 1 September 2026. After that date those users drop from 3,000 credits to 1,900, and from 7,000 to 3,900. That date is still in front of you. It is the next commercial event, not a historical footnote.
| Decision | Why it still changes the meeting |
|---|---|
| Who is on which plan, this week | Seat prices did not move. Plan mix still sets the floor and the included pool. |
| Credit burn versus allotment, by user and by model | The seat report can look clean while overflow is already the bill. |
| Premium-model mix | Token rates differ by model. A request count will not show that. |
| Overflow policy | Additional usage is on by default. Completions stay unlimited. That is a written policy, not a default you can ignore. |
| Existing Business and Enterprise grace | Extra credits run only through 1 September 2026. After that, the allotment falls. |
| Annual Pro and Pro+ riders still on term | They stay on request-based pricing until expiry, then drop to Copilot Free unless converted. |
Can a non-specialist brief the CIO in fifteen minutes from the pack you have today? If the answer is no, you do not have a Copilot position. You have a seat list.
The decision layer, not another inventory
You already have a GitHub admin export and a Microsoft seat file. The gap is not another inventory. The gap is turning that estate into a decision: which seats are real, which models are burning the pool, and what the 1 September drop does to overflow. LICENSEWARE sits on the inventory and ITSM tools you already run. It is not a rip-and-replace SAM suite. It is a decision layer: what matters, why it matters now, what should happen next. That is cost optimization on seats plus meters, not another named-user report.
If you are heading into a Microsoft conversation with a clean seat count and no credit position, book a Software Intelligence Review. You can also start on the free plan and run the analysis on your own data.
The question to walk in with
Do not let Microsoft frame GitHub Copilot credits as “the price stayed the same.” The right question is: what are we actually consuming, which models are doing the consuming, what happens when the grace allotment ends on 1 September, and what does the next term cost against this estate rather than against last year’s request cap. That is a data question, not a sales question.
The vendor will walk in with an unchanged SKU. The only question is whether you have the meter first: current, defensible, and tied to the contract in front of you.
FAQ
When did GitHub Copilot credits start? 1 June 2026. GitHub announced the change on 27 April 2026.
Did Copilot seat prices change? No. Pro remains $10 a month, Pro+ $39, Business $19 per user, Enterprise $39 per user. What changed is the unit inside the seat: AI Credits, overflow, and model rates.
How many GitHub AI Credits does each plan include? Pro 1,000. Pro+ 3,900. Business 1,900 per user. Enterprise 3,900 per user. Existing Business and Enterprise customers received 3,000 and 7,000 through 1 September 2026. Each credit is $0.01.
What stays unlimited after the credit pool is gone? Code completions and Next Edit Suggestions on paid plans. Chat and longer agent-style work stop unless you set an overflow budget.
