▲ Independent reporting on the agent economy  ·  editor@gaasnews.com
Sections
HomeWhat is GaaS?PlatformsOrchestrationProtocolsEnterprisePricingFundingLaborSafetyPolicyOpinionGlossaryAboutContact
Home›Pricing & Models›GitHub Copilot credits
Pricing & Models

GitHub now puts Copilot in every Enterprise seat, with no AI credits included

The October 28 changelog entry reads like a giveaway. It is closer to the end of the included allowance: the seat stops being what buys AI, and the enterprise budget takes over.

AJ
Andrew Jamerson
Founding Editor
Oct 31, 2026 · 4 min read
A monitor showing analytics charts and gaugesPhoto: A monitor showing analytics charts and gauges. // GaaS News
TL;DR
  • GitHub Enterprise customers with an account team can now get Copilot for every licensed user. This version "does not include monthly AI credits per seat."
  • Usage is "metered from the first request" against an enterprise budget. Copilot Business includes 1,900 credits a month and Copilot Enterprise 3,900, worth $19 and $39 at $0.01 a credit.
  • Code completion keeps working with no budget. Chat, code review and cloud agents stop until budget is available.

GitHub Copilot is now included with GitHub Enterprise for every licensed user at customers that work with a GitHub or Microsoft account team, the company said in an October 28 changelog entry. The catch sits a few lines down. This access "does not include monthly AI credits per seat," and "AI usage is metered from the first request against your enterprise budget." GitHub adds: "There is no included usage allowance to track or top up."

Compare that with the plans it sits beside. GitHub's billing documentation lists 1,900 AI credits per user per month for Copilot Business and 3,900 for Copilot Enterprise, pooled across the billing entity, with "1 AI credit = $0.01 USD." That makes the included usage worth $19 and $39 per user. Under the new arrangement it is zero.

How GitHub Copilot usage is billed now

GitHub Copilot usage is billed by tokens converted into AI credits. "The cost of an interaction depends on two things: the model and the number of tokens consumed," the documentation says. It draws the contrast directly: a quick chat question on a lightweight model "might cost a fraction of a credit," while "a long coding agent session using a frontier model across multiple files will cost more, because it's doing more work."

That sentence is the whole story of agent pricing in miniature. A completion costs almost nothing. An agent working through a repository for an hour can cost a lot, and how much depends on choices no procurement team sees. GitHub has drawn its line accordingly. Code completion and next edit suggestions "keep working even with no budget available." Copilot Chat, Copilot code review and Copilot in the cloud "require available budget to use." When the money runs out, the agent features stop and the autocomplete stays. "An exhausted budget never removes access," GitHub says, meaning the seat survives.

Who gets the GitHub Copilot seat

The GitHub Copilot seat in this form is not for everyone yet. It is "available now for GitHub Enterprise Cloud" through account teams. For self-serve customers it is "currently unavailable," with dates promised "in the coming weeks." People in an enterprise account without a GitHub Enterprise license can get the same access through Copilot Standalone at $10 per user per month, also with no included credits. Copilot Business and Copilot Enterprise remain on sale.

GitHub has not published what the bundled tier adds to the price of a GitHub Enterprise seat. That number is what buyers need most, and sales teams will quote it individually.

Why the seat stopped being the unit

The seat stopped being the unit because agents broke the math behind it. GitHub says as much: "Previously, you had to manage two things: how many Copilot seats you were paying for and how much AI usage those users consumed." Now "usage against your enterprise budget is the primary thing to manage." A flat seat works when usage per person falls in a narrow range. Coding agents widened that range, so the vendor is moving the variable cost onto the customer's budget and keeping the seat as access.

Microsoft's own numbers point the same way. On its October 28 earnings call, the company described usage-based offerings as "primarily GitHub Copilot," and it declined to update its Microsoft 365 Copilot seat count on the same day. Other vendors are on similar paths. Cursor's billing has been in flux for months, and outcome-based pricing is the customer-service version of the same shift.

What the GitHub change does not establish

What the GitHub change does not establish is whether customers pay more or less. Without the bundled price, nobody outside a sales conversation can run the comparison. The changelog does not say whether existing Copilot Enterprise customers will be pushed to the new model. It says current self-serve subscriptions are unaffected.

What engineering leaders should check

Engineering leaders should model this before the sales call. Take last quarter's credit use by team, split agent sessions from chat, and work out what each would cost with no allowance. At the Copilot Enterprise rate, 100 developers carry 390,000 included credits a month, or $3,900 of usage that would need to appear somewhere else. Set budgets and alerts by team first, because under this model the budget is the control. Cheap tokens do not make cheap agents, and that is now a line on the invoice.

AJ

Andrew Jamerson

Founding Editor, GaaS News

Andrew Jamerson is the founding editor of GaaS News, covering the economics of the agent era. He started the publication to cover Agentic AI as a Service as a dedicated beat and edits every article on the site.

Be on the list when the beat breaks

One email when a platform ships, a round closes, or the ground shifts under the software stack.