On 28 October 2026 GitHub announced that Copilot Fusion is generally available. Rather than picking a named model, a developer picks Fusion and an effort level (Standard, High or Max), and Copilot decides which models to use for each prompt and how they work together.
How it works
GitHub describes three patterns. Single: one model handles the task. Cascade: an efficient model drafts, a quality check accepts the result or escalates to a stronger model. Critique: one model drafts, another reviews, and the first revises.
The claim, and its source
GitHub says Fusion Max scored 60 on the Artificial Analysis Coding Agent Index, within two points of Claude Code with Opus 5.5 Max, at about half the API cost per task, roughly half the tokens and about 40% less time. That is GitHub’s own benchmark claim, not an independent result.
Availability and billing
Fusion is on all Copilot plans in VS Code, the Copilot app, the Copilot CLI and JetBrains IDEs. Users are billed for each model Fusion uses at that model’s standard rate, and paid plans get a 30% discount on model costs for a limited time. Admins can restrict which models Fusion may use and set a default effort level through enterprise managed settings.
What it means
The model is no longer the product. Tools now choose models for the user, so the cost of an AI-assisted build depends on routing and effort settings that most buyers never see. It also means “which AI model do you use?” is becoming the wrong question to put to a supplier.
So what
Ask suppliers how they control AI cost and quality, not which logo they use: who sets effort levels, who reviews output, and whether savings reach your invoice. Our AI-assisted development work keeps a named engineer accountable for every change whatever the tool routes to. To discuss a build, start a project.