Software Engineering Tasks on SWE-rebench subset V2 (test)
43.7Resolved RateClaude Opus 4.6
Evaluation Results
| Method | Links | |
|---|---|---|
| Claude Opus 4.6Scaffold=Claude Code2026.03 | 43.7 | |
| KAT-Coder-V2Scaffold=Claude Code2026.03 | 43.3 | |
| KAT-Coder-V2Scaffold=OpenCode2026.03 | 38.7 | |
| Claude Opus 4.6Scaffold=OpenCode2026.03 | 37.3 |