Cognition launches Devin Fusion: first multi-model coding agent hits leaderboard near Claude Code

On Sept 12, Cognition released Devin Fusion, a coding agent harness billed as the most efficient frontier coding agent, pairing a frontier main model (e.g., Claude Fable 5.1 or GPT-6 Ast) with a low-cost sidekick: strong models for planning, cheaper models for execution. Cognition also shipped Fusion in Devin CLI, claiming it is 39% cheaper than prior setups on coding benchmarks.

Confirmed

• Artificial Analysis ran an independent same-day eval, making Devin Fusion the first multi-model coding agent to enter its Coding Agent Index (v1.5).
• The Fable 5.1 (xhigh) + SWE-2 (medium) config scored 61.7, saving 36% per task while trailing Claude Code by only 4%.
• Full per-category results for two configurations were published, confirming strong performance at significantly lower cost.

Why it matters

• Devin Fusion is the first proof that a multi-model architecture (strong planner + cheap executor) can compete with top-tier coding agents, undercutting Claude Code on cost with only a small performance gap.

2026-09-12 ~ 2026-09-12 · 6 related posts

• [source] Cognition's Devin CLI Fusion Cuts Coding Benchmark Costs 39% by Splitting Planner and Executor — DeryaTR_ · 2026-09-12
• [source] Devin Fusion benchmarks: frontier lead model plus cheap sidekick cuts coding agent costs — ArtificialAnlys · 2026-09-12
• Devin Fusion scores 61.7 on Coding Agent Index, nearly matching Claude Code at 36% less cost — ArtificialAnlys · 2026-09-12
• [source] Devin Fusion hits 61.7 on Coding Agent Index, costs 36% less than Claude Code per task — ArtificialAnlys · 2026-09-12
• Devin Fusion Debuts: Multi-Model Coding Agent Scores 62, First on AA Index — AccBalanced · 2026-09-12

1 near-duplicate retellings: daniel_mac8