GitHub Copilot Fusion is generally available for multi-model routing
GitHub made Copilot Fusion generally available: developers pick Fusion in the model picker and let it route and orchestrate models per task, with Standard, High, and Max effort levels. GitHub says Fusion Max scored 60 on the Artificial Analysis Coding Agent Index. Usage is billed at each underlying model’s rate, with a limited-time 30% model-cost discount on paid Copilot plans.
On 28 October 2026 GitHub’s Changelog announced that GitHub Copilot Fusion is generally available. Choosing Copilot Fusion in the model picker means you are not locking a single model; Fusion selects and orchestrates models for each prompt while you pick an effort level—Fusion Standard, High, or Max. Workflow modes include Single, Cascade (draft then quality-gate escalate), and Critique (draft, review, revise).
GitHub says Fusion combines Project HydraFusion multi-model orchestration with Copilot auto model selection. On the Artificial Analysis Coding Agent Index, Copilot Fusion Max scored 60, within about two points of Claude Code with Opus 5.5 Max, at roughly half the average API cost per task, about half the tokens, and about 40% less time per task, per GitHub. Admins control which models Fusion may use and whether Fusion is the default; Fusion only uses models enabled by enterprise/organization policy.
Billing charges each model Fusion invokes at that model’s standard rate; paid Copilot plans receive a limited-time 30% discount on model costs when using Fusion. GitHub says Fusion is available on all Copilot plans in Visual Studio Code, the GitHub Copilot app, GitHub Copilot CLI, and JetBrains IDEs, with more clients coming. Preview users of auto, HydraFusion, or HydraFusion Max are mapped to the matching Fusion effort levels.