List your company

Software Engineering Issue Solving on SWE-bench Verified (Pass@K)

51.2Pass@1HHD
37.47241.03644.648.164May 12, 2026

Evaluation results

MethodLinks
HHDBackbone Model=Qwen-72b2026.0551.26770.2
HHD-MBackbone Model=Qwen-72b2026.0547.261.867.8
Dense Expert JudgeBackbone Model=Qwen-72b2026.0545.86568.8
Agent-RLVR-PlanBackbone Model=Qwen-72b2026.0545.665.470.2
Naive RFTBackbone Model=Qwen-72b2026.054462.267.2
SE-agent-ReflectBackbone Model=Qwen-72b2026.0543.464.469
Agent-RLVR-EFBackbone Model=Qwen-72b2026.0543.46569
HHDBackbone Model=GLM-Air2026.054255.460.8
HHD-MBackbone Model=GLM-Air2026.054154.659.6
SE-agent-ReflectBackbone Model=GLM-Air2026.054055.259.8
Agent-RLVR-EFBackbone Model=GLM-Air2026.0539.454.459.6
Dense Expert JudgeBackbone Model=GLM-Air2026.0539.455.259
Agent-RLVR-PlanBackbone Model=GLM-Air2026.0538.65560.2
Naive RFTBackbone Model=GLM-Air2026.05385358.8