List your company

Agent Safety Judgment on ROME SDM unsafe

100F1 ScoreClaude Sonnet 4.6
92.77294.648596.52598.4015May 5, 2026

Evaluation results

MethodLinks
Claude Sonnet 4.6Model=Claude Sonnet 4.62026.0510010097
GPT-5.2Model=GPT-5.22026.0598.4897100
Claude Opus 4.6Model=Claude Opus 4.62026.0598.4396.9197
GPT-5Model=GPT-52026.0593.0587100