Artificial Analysis benchmarks GitHub Copilot Fusion against competing coding agents

What happened

Artificial Analysis benchmarked GitHub Copilot Fusion’s max setting in GitHub Copilot CLI on its Coding Agent Index v2.0, finding it scores 2 points below Opus 5.5 in Claude Code at roughly half the Cost per Task. Copilot Fusion costs $5.59 per task versus $11.42 for Claude Code with Opus 5.5 (max), and uses 7.2M tokens per task. It leads Terminal-Bench 4.0 with 71, but ranks third on FrontierCode with 51 and scores 57 on SWE-Atlas-QnA-AA.

Read next

• More changes in today’s briefingGoogle DeepMind announces Genie 4 for real-time world simulation with native audio

The story so far

Artificial Analysis benchmarked GitHub Copilot Fusion’s max setting in GitHub Copilot CLI and found it scored 2 points below Opus 5.5 (max) in Claude Code on its Coding Agent Index v2.0, at roughly half the cost per task. The measured cost was $5.59 per task for Copilot Fusion versus $11.42 for Claude Code with Opus 5.5 (max); Fusion used 7.2 million tokens per task. The index weights three benchmarks equally. Artificial Analysis reported that Fusion led Terminal-Bench 4.0 with a score of 71, ranked third on FrontierCode with 51, and scored 57 on SWE-Atlas-QnA-AA. Token costs are charged according to the underlying models used to complete each request.