GitHub Copilot Fusion updates: more availability, benchmarks, and models
GitHub has expanded its Copilot Fusion feature with new model additions, broader availability across web, mobile, and Visual Studio, and performance results on Terminal-Bench 4.1 public leaderboards. According to the company, Copilot Fusion Max achieved a 71.2% resolution rate on the benchmark while undercutting competing frontier models on cost. Open article →
- GitHub Brings Copilot Fusion Model Orchestration to Visual Studio
- terminal-bench/terminal-bench | Harbor
- The best models are no longer single models - they are fusions of multiple frontier models, tight...
- GitHub Copilot Fusion now supports more models and platforms, with top performance on public Term...