List your company

Jailbreak Detection on JailbreakBench AutoDAN-T

76Detection RateGuardReasoner
15.6831.344762.66May 11, 2026

Evaluation results

MethodLinks
GuardReasonerTarget Model=Vicuna2026.0576
GuardReasonerTarget Model=Llama22026.0568
embedding disruption methodTarget Model=Llama22026.0568
GuardReasonerTarget Model=Qwen2.52026.0565
embedding disruption methodTarget Model=Vicuna2026.0565
embedding disruption methodTarget Model=Qwen2.52026.0552
Llama-Guard-3Target Model=Vicuna2026.0549
Llama-Guard-3Target Model=Llama22026.0532
Llama-Guard-3Target Model=Qwen2.52026.0518