List your company

Software Engineering Tasks on SWE-bench Multilingual (test)

77.8Resolution Rate (%)Claude Opus 4.6
69.89671.9487476.052Mar 29, 2026

Evaluation results

MethodLinks
Claude Opus 4.6Scaffold=Claude Code2026.0377.8
KAT-Coder-V2Scaffold=Claude Code2026.0375.4
KAT-Coder-V2Scaffold=OpenCode2026.0371.2
Claude Opus 4.6Scaffold=OpenCode2026.0370.2