List your company

Jailbreak Attack on AdvBench (ASR)

32Attack Success Rate (ASR)STEER
14.00818.67923.3528.021Jul 2, 2026

Evaluation results

MethodLinks
STEERTarget Model=GPT-4o-mini2026.0732
GCGTarget Model=GPT-4o-mini2026.0726.9
CSRTTarget Model=GPT-4o-mini2026.0714.7