List your company

Software Engineering Task Resolution on SWE-bench In-Distribution

52.34Resolved RateSkillFlow
14.972824.673934.37544.0761May 13, 2026

Evaluation results

MethodLinks
SkillFlowMethod Category=Ours2026.0552.34
FlowSteerMethod Category=Agent+RL2026.0543.75
AgentFlowMethod Category=Agent+RL2026.0541.41
SkillRLMethod Category=Agent+RL2026.0539.06
v4-flashMethod Category=Baseline2026.0538.28
Qwen3.5-AFlowMethod Category=AFlow, Backbone=Qwen3.52026.0530.47
Qwen3.5-GRPOMethod Category=GRPO, Backbone=Qwen3.52026.0518.75
Qwen3.5-9BMethod Category=Baseline2026.0517.19
Qwen3.5-SFTMethod Category=SFT, Backbone=Qwen3.52026.0516.41