Task Success on SkillLearnBench Random
29Success CountRevision v3
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Revision v3Model=GPT-5.5, Harness=Codex2026.05 | 29 | 18 | |
| Revision v3Model=Opus-4.7, Harness=Claude Code2026.05 | 25 | 40 | |
| Revision v3Model=DeepSeek-V4-Pro, Harness=Claude Code2026.05 | 24 | 24 | |
| Skill v0Model=GPT-5.5, Harness=Codex2026.05 | 23 | 6 | |
| Revision v3Model=Kimi-2.6, Harness=Claude Code2026.05 | 22 | 32 | |
| No skillModel=GPT-5.5, Harness=Codex2026.05 | 20 | — | |
| Skill-CreatorModel=GPT-5.5, Harness=Codex2026.05 | 17 | -6 | |
| Skill-CreatorModel=DeepSeek-V4-Pro, Harness=Claude Code2026.05 | 16 | 8 | |
| Revision v3Model=Qwen-3.6plus, Harness=Claude Code2026.05 | 15 | 20 | |
| Skill v0Model=DeepSeek-V4-Pro, Harness=Claude Code2026.05 | 14 | 4 | |
| No skillModel=DeepSeek-V4-Pro, Harness=Claude Code2026.05 | 12 | — | |
| Skill v0Model=Kimi-2.6, Harness=Claude Code2026.05 | 11 | 14.83 | |
| Skill-CreatorModel=Kimi-2.6, Harness=Claude Code2026.05 | 10 | 8 | |
| Skill-CreatorModel=Qwen-3.6plus, Harness=Claude Code2026.05 | 8 | 6 | |
| No skillModel=Opus-4.7, Harness=Claude Code2026.05 | 7 | — | |
| Skill-CreatorModel=Opus-4.7, Harness=Claude Code2026.05 | 7 | 0 | |
| Skill v0Model=Opus-4.7, Harness=Claude Code2026.05 | 7 | 0 | |
| Skill v0Model=Qwen-3.6plus, Harness=Claude Code2026.05 | 7 | 4 | |
| No skillModel=Kimi-2.6, Harness=Claude Code2026.05 | 6 | — | |
| No skillModel=Qwen-3.6plus, Harness=Claude Code2026.05 | 5 | — |