Companies on GitHub Enterprise Cloud can now give every licensed user GitHub Copilot without buying a separate Copilot seat for each of them. They pay for the AI usage instead. GitHub announced the change at its Universe developer conference in San Francisco, along with a new Copilot app, a model router called Copilot Fusion and a deeper tier of automated code review.
For now the offer is limited to customers who work with a GitHub or Microsoft account team. Self-serve customers will get it in stages, and GitHub said in a changelog post that it will share dates “in the coming weeks.” Copilot Business and Copilot Enterprise stay on sale, and existing subscriptions don’t change.
How the bill works now
Under the new arrangement, usage is metered from the first request against the enterprise’s AI budget, with no monthly credit allowance attached to each seat. That differs from the current plans. Copilot Business costs $19 per seat a month and Copilot Enterprise $39, and each seat adds 1,900 or 3,900 AI credits to a shared pool, with anything beyond the pool billed at $0.01 per credit, GitHub’s plan documentation says.
Running out of budget doesn’t take away the seat. Code completion and next edit suggestions keep working with no budget left, while Copilot Chat, Copilot code review and Copilot in the cloud need money in the budget to run. “An exhausted budget never removes access,” GitHub wrote. People in an enterprise account who don’t hold a GitHub Enterprise license can get the same Copilot features through Copilot Standalone at $10 per user per month.
The app, the router and a heavier review
Jay Parikh, Microsoft’s executive vice president of CoreAI, whose organization includes GitHub, framed the launches around agents in a blog post. “In a couple of years, we’ve gone from writing every line of code to working with AI assistants to orchestrating fleets of agents,” he wrote.
The GitHub Copilot app is built to carry work from an issue to a merged change in one place, and it can also be used for specs, design discussions and telemetry analysis. Copilot Fusion decides how to run a task across one or more models. GitHub said Fusion Max scored within two points of Opus 5.5 Max on the Artificial Analysis Coding Agent Index at half the cost, taking 40% less time per task. Copilot projects, available in the app starting today, run agent sessions across several repositories in parallel. A cloud agent session can also now follow a developer between the app, VS Code, the command line and the web.
On the review side, GitHub added Copilot code review Max, which it said posted the highest F1 score on ReviewBench for high-severity issues. The existing Lite and Balanced settings remain for everyday reviews. On Monday, GitHub also let a model decide which Copilot commands can run without asking, part of a run of releases aimed at letting agents do more on their own.
GitHub’s own usage data, released the same day in its Octoverse report, shows how much of the work on the platform now starts with agents. More than 40% of pull requests on GitHub started with an agent’s first draft in September, up from 17% in March. In the same month, 61% of Copilot users who picked a model themselves used more than one.
That load is landing on GitHub’s servers too. The company said it has added 4 million CPU cores, 160 PB of high-speed storage and 50 Tb/s of network capacity so far this year, and it expects to finish moving to a multi-region deployment in Azure early next year.