GitHub Releases Copilot Fusion

GitHub released Copilot Fusion in GitHub Copilot on October 28, 2026. The system orchestrates models from Anthropic and OpenAI; in the Coding Agent Index, it scored 2 points below Claude Opus 5.5 in Claude Code while costing half as much per task. GitHub describes Fusion as a runtime orchestration system that selects and coordinates models to balance performance, cost and latency for each task. Its options include Claude Opus 5.5, Claude Sonnet 5.5, GPT-6 Astra and GPT-6 Sol. Artificial Analysis benchmarked the max setting in GitHub Copilot CLI for Coding Agent Index v2.0. Token charges depend on the underlying models used for each request.

Copilot Fusion cost $5.59 per task, compared with $11.42 for Claude Code with Opus 5.5 (max). Devin Fusion with Opus 5.5 (max) and GPT-6 Luna (high) scored 62 at $8.00 per task. The index gives equal weight to its three benchmarks. Copilot Fusion led Terminal-Bench 4.0 with 71, ranked third on FrontierCode with 51, behind Claude Code at 56 and Devin Fusion at 54, and scored 57 on SWE-Atlas-QnA-AA, between GPT-6 Astra in Codex and Gemini 4 Argon in Antigravity CLI.

Copilot Fusion used 7.2M tokens per task, approximately half the amount used by Claude Opus 5.5 in Claude Code and 30% less than Devin Fusion with Opus 5.5 (max) and GPT-6 Luna (high).

GitHub released Copilot Fusion in GitHub Copilot on October 28, 2026. The system orchestrates models from Anthropic and OpenAI; in the Coding Agent Index, it scored 2 points below Claude Opus 5.5 in Claude Code while costing half as much per task. GitHub describes Fusion as a runtime orchestration system that selects and coordinates models to balance performance, cost and latency for each task. Its options include Claude Opus 5.5, Claude Sonnet 5.5, GPT-6 Astra and GPT-6 Sol. Artificial Analysis benchmarked the max setting in GitHub Copilot CLI for Coding Agent Index v2.0. Token charges depend on the underlying models used for each request.

Copilot Fusion cost $5.59 per task, compared with $11.42 for Claude Code with Opus 5.5 (max). Devin Fusion with Opus 5.5 (max) and GPT-6 Luna (high) scored 62 at $8.00 per task. The index gives equal weight to its three benchmarks. Copilot Fusion led Terminal-Bench 4.0 with 71, ranked third on FrontierCode with 51, behind Claude Code at 56 and Devin Fusion at 54, and scored 57 on SWE-Atlas-QnA-AA, between GPT-6 Astra in Codex and Gemini 4 Argon in Antigravity CLI.

Copilot Fusion used 7.2M tokens per task, approximately half the amount used by Claude Opus 5.5 in Claude Code and 30% less than Devin Fusion with Opus 5.5 (max) and GPT-6 Luna (high).