List your company

Software Engineering Task Success on SWE-Skills-Bench Hard

35Task Success RateRevision v3
19.423.4527.531.55May 31, 2026

Evaluation results

MethodLinks
Revision v3Model=Qwen-3.6plus, Harness=Claude Code2026.053518.57
Revision v3Model=GPT-5.5, Harness=Codex2026.05337.14
No skillModel=Opus-4.7, Harness=Claude Code2026.0532—
Revision v3Model=Opus-4.7, Harness=Claude Code2026.0531-1.43
Revision v3Model=DeepSeek-V4-Pro, Harness=Claude Code2026.053010
Skill v0Model=GPT-5.5, Harness=Codex2026.05291.43
No skillModel=GPT-5.5, Harness=Codex2026.0528—
Skill v0Model=DeepSeek-V4-Pro, Harness=Claude Code2026.05287.14
Skill-CreatorModel=GPT-5.5, Harness=Codex2026.0527-1.43
Revision v3Model=Kimi-2.6, Harness=Claude Code2026.05272.86
Skill v0Model=Qwen-3.6plus, Harness=Claude Code2026.05266.25
Skill-CreatorModel=Opus-4.7, Harness=Claude Code2026.0525-10
No skillModel=Kimi-2.6, Harness=Claude Code2026.0525—
Skill-CreatorModel=DeepSeek-V4-Pro, Harness=Claude Code2026.05252.86
Skill-CreatorModel=Qwen-3.6plus, Harness=Claude Code2026.05242.86
Skill v0Model=Opus-4.7, Harness=Claude Code2026.0523-12.86
Skill v0Model=Kimi-2.6, Harness=Claude Code2026.05236.88
No skillModel=DeepSeek-V4-Pro, Harness=Claude Code2026.0523—
No skillModel=Qwen-3.6plus, Harness=Claude Code2026.0522—
Skill-CreatorModel=Kimi-2.6, Harness=Claude Code2026.0520-7.14