List your company

Instruction-following and procedural reasoning on SOP-Bench

100AccuracyGA
84.488.4592.596.55Apr 18, 2026

Evaluation results

MethodLinks
GAModel=Claude Sonnet 4.62026.041002.02532.080.48
OpenClawModel=Claude Sonnet 4.62026.041002.6402.640.38
OpenClawModel=Minimax M2.72026.04952.91462.960.32
GAModel=Minimax M2.72026.0490893329240.97
Claude CodeModel=Claude Sonnet 4.62026.04851.23231.250.68