List your company

Agentic Coding on Multi-SWE-Bench

44.3Pass@1Claude Sonnet-4.5
19.75626.12832.538.872Mar 21, 2026

Evaluation results

MethodLinks
Claude Sonnet-4.52026.0344.3
Gemini 3-pro2026.0342.7
Seed1.82026.0342
GPT-5 High2026.0341.7
Gemini 2.5-pro2026.0320.7