Artificial Analysis Benchmarks GitHub Copilot Fusion Two Points Below Claude Code With Opus 5.5 at Half the Cost per Task
GitHub has released Copilot Fusion in GitHub Copilot, which orchestrates between Anthropic and OpenAI models. It scores 2 points below Opus 5.5 in Claude Code at half the Cost per Task in the Coding Agent Index GitHub describes Copilot Fusion as a runtime orchestration system, which selects and coordinates models to optimize performance, cost, and latency for each task. Copilot Fusion chooses between a number of models from Anthropic and OpenAI, including Claude Opus 5.5, Claude Sonnet 5.5, GPT-6 Astra, and GPT-6 Sol. We benchmarked its max setting in GitHub Copilot CLI in the Artificial Analysis Coding Agent Index v2.0. Token costs are charged based on the underlying models used to complete a request. Key takeaways: ➤ Cost-efficient for near-frontier performance: Copilot Fusion in GitHub Copilot costs $5.59 per task, versus $11.42 for Claude Code with Opus 5.5 (max). Devin Fusion with Opus 5.5 (max) + GPT-6 Luna (high) scores 62 at $8.00 per task. ➤ Slight lead in Terminal-Bench 4.0: The Coding Agent Index v2.0 weights its three constituent benchmarks equally. Copilot Fusion in GitHub Copilot leads Terminal-Bench 4.0 with a score of 71. It ranks third on FrontierCode in the chart at 51, behind Claude Code with Opus 5.5 (max) at 56 and Devin Fusion with Opus 5.5 (max) + GPT-6 Luna (high) at 54. Its SWE-Atlas-QnA-AA score lags at 57, between GPT-6 Astra in Codex and Gemini 4 Argon in Antigravity CLI. ➤ Half the token use of Opus 5.5 in Claude Code: GitHub Copilot Fusion used 7.2M tokens per task in the Coding Agent Index, approximately half of the tokens used by Claude Opus 5.5 in Claude Code, and 30% less than Devin Fusion with Opus 5.5 (max) + GPT-6 Luna (high).
Context
It suggests routing each task between Anthropic and OpenAI models can deliver near frontier coding agent results for much less money, which matters for teams paying per token.
GitHub Copilot is GitHub's AI coding assistant. Copilot Fusion is a new mode that GitHub describes as a runtime orchestration system, choosing among models from Anthropic and OpenAI for each task. Artificial Analysis is an independent benchmarking firm; its Coding Agent Index v2.0 combines FrontierCode, Terminal Bench 4.0 and SWE Atlas QnA.
Deep dive
Artificial Analysis said on Wednesday that GitHub has released Copilot Fusion in GitHub Copilot, which orchestrates between Anthropic and OpenAI models. According to Artificial Analysis, Fusion chooses among models including Claude Opus 5.5, Claude Sonnet 5.5, GPT 6 Astra and GPT 6 Sol, and token costs are charged based on the underlying models used.
Artificial Analysis benchmarked the max setting in GitHub Copilot CLI. It reports a Coding Agent Index score of 60, two points below Claude Code with Opus 5.5 at max, at 5.59 dollars per task versus 11.42 dollars. Devin Fusion, which pairs Opus 5.5 with GPT 6 Luna, scored 62 at 8.00 dollars per task.
By benchmark, Artificial Analysis says Copilot Fusion leads Terminal Bench 4.0 with 71, ranks third on FrontierCode with 51, and lags on SWE Atlas QnA with 57. It also reports that Fusion used 7.2 million tokens per task, about half of Opus 5.5 in Claude Code and 30 percent less than Devin Fusion.
These are Artificial Analysis results on its own index. GitHub's own figures were not included in the post.
What to expect next
Watch whether GitHub publishes its own numbers and pricing details for Fusion, and whether the cost advantage holds as underlying model prices change. Multi model routing is becoming a competitive feature, with Devin Fusion already in the same chart, so other coding tools may follow. Independent tests on real repositories, beyond benchmark suites, will show whether the savings come with tradeoffs in quality.