List your company

Skill Generation on SkillsBench

22Baseline Success RateAutoRefine
16.69618.07319.4520.827May 9, 2026

Evaluation results

MethodLinks
AutoRefineTeacher Model=claude-sonnet-4.6, Teacher Trajectories=SPARK teacher trajectories2026.052226.140.743.418.617.2199
SPARKTeacher Model=claude-sonnet-4.6, Teacher Trajectories=SPARK teacher trajectories2026.05222450.853.128.829.1213
EvoSkillTeacher Model=claude-sonnet-4.6, Teacher Trajectories=SPARK teacher trajectories2026.0520.324.630.535.310.210.7112
Trace2SkillTeacher Model=claude-sonnet-4.6, Teacher Trajectories=SPARK teacher trajectories2026.0516.921.827.134.810.213124