For two years the first question about Copilot cost was "how many seats?" As of October 28, for enterprises that take the new model, that question costs nothing to get wrong. The seat is $0. What you pay for is what people actually use, from the first request, with no allowance in front of it.
That's mostly good news. It's also a change in where the risk sits, and the risk moves somewhere most teams aren't watching yet.
What GitHub shipped
On October 28, 2026, GitHub announced that enterprises working with a GitHub or Microsoft account team can get Copilot included with GitHub Enterprise for every licensed user (GitHub Changelog, October 2026). GitHub calls this "Copilot access". It has the same features and models as Copilot Enterprise, but it "does not include monthly AI credits per seat".
The pieces, from the changelog and GitHub's transition guide (GitHub Docs, What's changing with GitHub Enterprise and Copilot, retrieved 2026-11-03):
| Seat charge | Included AI credits | Usage | |
|---|---|---|---|
| Copilot access, GitHub Enterprise-licensed user | $0 | none | metered from first use |
| Copilot Standalone, user without an Enterprise license | $10/user/month | none | metered from first use |
| Copilot Business | $19/user/month | 1,900/user/month, pooled | overage beyond the pool |
| Copilot Enterprise | $39/user/month | 3,900/user/month, pooled | overage beyond the pool |
One AI credit is $0.01, so Business's 1,900 credits are worth exactly $19 and Enterprise's 3,900 exactly $39. The allowance is the seat price. Both allowances are pooled at the billing entity level and reset on the first of the month at 00:00 UTC (GitHub Docs, Copilot billing for organizations and enterprises, retrieved 2026-11-03).
Three availability facts matter before any of the math does:
- It's account-team only for now. Self-serve customers can't get it yet. GitHub says it will open in stages with dates "in the coming weeks". GitHub Free, GitHub Team and self-serve Enterprise Cloud trials aren't eligible.
- It's opt-in. Business and Enterprise customers don't transition automatically, and a transition takes effect on the first day of a month.
- It doesn't mix. You can't combine Copilot access with Business or Enterprise seats in the same enterprise, not even in separate organizations inside it. It's a whole-enterprise decision.
When is the free seat actually cheaper?
On list prices, for a GitHub Enterprise-licensed user, Copilot access is never more expensive than a Business seat. The gap is exactly the part of the pooled allowance you didn't use.
The arithmetic is short. A Business enterprise of N users pays 19 × N for the pool, plus anything consumed beyond it. A Copilot access enterprise pays only what it consumes. If average consumption is below 1,900 credits a user, Business pays for credits nobody spent. Above 1,900, the two bills are identical, because a Business enterprise is paying overage at the same per-credit rate.
Put real numbers on it. A 100-person enterprise averaging 800 credits a user pays $1,900 a month on Business and $800 on Copilot access. On Copilot Enterprise, which has the same feature set Copilot access now matches, it pays $3,900. Even an enterprise averaging 2,500 credits a user pays $2,500 either way on Business, and still $3,900 on Enterprise.
Standalone is the one that can lose. For a user without an Enterprise license it's $10 plus usage, against Business's $19 with $19 of usage included. Standalone wins below 900 credits a month, Business wins above it, and from 1,900 credits up Standalone costs exactly $10 a user more.
So where did the risk go?
The seat used to be the floor and the visible number. A Business invoice started at seats × $19, and the pool absorbed ordinary variation underneath it. Copilot access takes the floor away. GitHub's own framing is that "usage against your enterprise budget is the primary thing to manage", and that's exactly right. It's now the only thing.
GitHub's documentation is specific about what keeps working when there's no budget left. Inline code completions and next edit suggestions "are not billed in AI credits and remain unlimited". Copilot Chat, Copilot code review and Copilot in the cloud need available budget. And "an exhausted budget never removes access". The seat and the free features stay.
Read that list again from a finance seat. The things that stop when budget runs out are chat, agents and review. Those are the features that drive both value and spend. A budget that's too tight doesn't save money quietly. It turns off code review in the middle of a release week.
And a budget that's too loose, or missing, is worse. The transition guide tells you to "set spending controls based on your cost estimate" before transitioning. It doesn't describe any default cap. We'd assume there isn't one until you've checked your own tenant.
Our finding: this release quietly invalidates the most popular Copilot savings recommendation, including ours. FinOpsAid's seat inventory flags Copilot seats with no activity in the last 30 days and prices each one at a default $19 a month. That's the Business list price, and it's what produces the "reclaim idle seats" dollar figure on our Savings page. For a GitHub Enterprise-licensed user on Copilot access, an idle seat costs $0. Reclaiming it saves nothing. If your enterprise transitions, treat our idle-seat savings number as not applicable until we read the plan type properly. Idle seats still matter for access hygiene (see dormant seats as a security cost), but under this model the money is in the busiest users, not the quietest.
That inversion is the real story. Under seat pricing, waste lives at the bottom of the usage distribution. Under metered pricing, it lives at the top: the handful of users running long agent sessions on the most expensive model. Our earlier piece on finding and reclaiming idle Copilot seats is still right for Business and Enterprise. It's the wrong first move here.
What to check before you ask your account team
- Price your current pool utilisation. Total AI credits consumed last month divided by your included pool is the whole decision. At 40% utilisation, Copilot access saves you 60% of the seat line. At 100% or more it saves nothing. Reconciling AI credits to your bill walks through getting that number.
- Count who lacks an Enterprise license. Those users go to Standalone at $10 plus usage. If they're light users, that's cheaper than Business. If they're heavy, it's $10 more each.
- Write the budget before the transition date, not after. Transitions start on the first of a month. Set the enterprise budget, then a cost-center layer, then per-user budgets for the heaviest consumers. User-level budgets are the only scope GitHub always hard-stops (what changes when user budgets expire).
- Decide which features you'd rather lose first. When budget runs out, chat, code review and cloud agents stop together. If code review is load-bearing in your merge process, give it headroom before you give chat headroom.
- Keep the "AI credits paid usage" policy on. GitHub says it's on by default and required for credit-consuming features. Turning it off to "save money" turns those features off.
- Check the firewall allowlist. If you only allow
*.business.githubcopilot.com, GitHub says to add*.enterprise.githubcopilot.comor access breaks at transition. - Remember it's all or nothing per enterprise. You can't pilot Copilot access in one org while the rest stays on Business seats.
Where FinOpsAid fits
FinOpsAid reads your GitHub org read-only and shows AI credit consumption per user, per team and per cost center, next to Actions, storage and Codespaces. Under Copilot access that per-user view becomes the main thing to watch, because consumption is now the whole bill. Three honest limits. The per-user AI credit figure GitHub exposes is a single daily number with no model or feature breakdown. GitHub's Copilot metrics lag roughly two to three days, so nothing here is real-time. And as noted above, our idle-seat savings figure assumes a paid seat. For an actual spending stop, use GitHub's own budgets. Related: cost centers and AI credit pools, budgets that catch overspend early, and what actually consumes GitHub AI credits.
Frequently asked questions
Does GitHub Enterprise include Copilot for free now?
The seat is free: a GitHub Enterprise-licensed user gets Copilot access at a $0 seat charge, with the same features and models as Copilot Enterprise. Usage isn't free. There are no included AI credits, and chat, code review and cloud agents are metered from the first request at $0.01 per credit. Code completions and next edit suggestions stay unlimited.
Is Copilot access cheaper than Copilot Business?
On list prices, for GitHub Enterprise-licensed users, it's never more expensive. Business charges $19 a user for a pooled 1,900-credit allowance. Copilot access charges only what's consumed. Below 1,900 credits per user on average, Copilot access is cheaper by the unused allowance. Above it, the bills are identical. Negotiated pricing can change that comparison.
What happens when the Copilot budget runs out under Copilot access?
GitHub says an exhausted budget never removes access. Code completions and next edit suggestions keep working. Copilot Chat, Copilot code review and Copilot in the cloud stop until budget is available again. So budget sizing decides which AI features your developers lose mid-month, not whether they keep their seat.
Sources
- GitHub, "GitHub Enterprise now includes access to GitHub Copilot", retrieved 2026-11-03, https://github.blog/changelog/2026-10-28-github-enterprise-now-includes-access-to-github-copilot
- GitHub Docs, "What's changing with GitHub Enterprise and Copilot", retrieved 2026-11-03, https://docs.github.com/copilot/get-started/whats-changing-with-github-enterprise-and-copilot
- GitHub Docs, "Copilot billing for organizations and enterprises", retrieved 2026-11-03, https://docs.github.com/copilot/concepts/billing-and-usage/organizations-and-enterprises/billing
If you want to know your real pool utilisation before you talk to your account team, connect your GitHub org — read-only and free during beta — or explore the demo dashboard first.