Skip to main content
Post

GitHub Copilot Fusion updates: more availability, benchmarks, and models

GitHub expanded Copilot Fusion. You can now pick it in the model picker on github.com, in Visual Studio, and in the GitHub Mobile beta apps for iOS and Android. Support in the production mobile apps is coming shortly.

Fusion isn't a single model. It's a runtime router that picks the models and execution pattern for each prompt, then returns one response. For example, it can have one model draft a solution while others critique it. You choose an effort level (Standard, High, or Max), and Fusion handles the rest.

This update also adds models to the pool:

  • Claude Sonnet 5.5 in Fusion Max
  • GPT-6.1 Terra in Fusion High
  • MAI-Code-2 in Fusion Standard

GitHub also says Fusion Max now appears on the public Terminal-Bench 4.1 leaderboards hosted by Stanford, Harbor, and the Laude Institute. By GitHub's numbers, it resolved 71.2% of tasks, 5.4 points ahead of Opus 5.5 and Fable 5.2. It also cost 57% less than Opus 5.5 and 88% less than Fable 5.2.

You're billed at each model's standard rate, with no separate Fusion charge. Fusion High and Max can use several models per task, so they may use more AI credits than Standard. Through November 30, 2026, paid Copilot plans get a 30% discount on model costs when using Fusion.

Loading replies