List your company

Skill-based attack evaluation on SkillHarm Self-Mutating Poisoning

69.3ASRGPT-5.4
7.00423.17739.3555.523Jun 1, 2026

Evaluation results

MethodLinks
GPT-5.4Harness=Codex2026.0669.377.41
GPT-5.5Harness=Codex2026.0665.672.73.1
Sonnet 4.6Harness=Claude Code2026.0651.670.22.1
Qwen-3.6 27BHarness=OpenCode2026.0651.6650
Gemini 3 FlashHarness=Gemini CLI2026.0645.861.40.5
Opus 4.7Harness=Claude Code2026.069.441.514.1