GitHub Copilot Fusion Updates Expand Model Choices and Availability
GitHub Copilot Fusion introduces new models, expanded availability across more platforms, and posts strong public benchmark results, with significant cost efficiency compared to leading alternatives.
What changed?
GitHub Copilot Fusion now supports more platforms, including the latest versions of github.com, Visual Studio, and GitHub Mobile beta apps for iOS and Android. The update introduces new models to different Fusion tiers: Claude Sonnet 5.5 on Fusion Max, GPT-6.1 Terra on Fusion High, and MAI-Code-2 on Fusion Standard. Users can now select Copilot Fusion in the model picker, letting it coordinate and select models based on the task and desired effort level. Additionally, Copilot Fusion is listed on the Terminal-Bench 4.1 leaderboards hosted by Stanford, Harbor, and the Laude Institute.
Why does it matter to an everyday developer?
These changes bring Copilot Fusion's advanced coding assistance to more environments, including web, desktop, and mobile. Developers can take advantage of leading-edge models for code generation directly where they work. Public benchmark results show Copilot Fusion Max achieving a 71.2% resolution rate on Terminal-Bench 4.1: 5.4 points ahead of prominent competitors, while costing considerably less. This means developers have access to top-performing AI coding assistance with measurable efficiency and cost-effectiveness.
What can the developer do now?
- Update to the latest version of github.com, Visual Studio, or GitHub Mobile beta to access Copilot Fusion.
- Select Copilot Fusion via the model picker to let the tool automatically coordinate among available models.
- Adjust the 'effort level' to fine-tune Fusion's approach to coding tasks.

