List your company

Multi-turn Jailbreaking on StrongReject (test)

0.34ASRSiren
0.31640.47570.6350.7943Jan 9, 2026

Evaluation results

MethodLinks
SirenTarget Model=Llama 3.3 70B Instruct2026.010.340.0197
SirenTarget Model=Average2026.010.610.0312
SirenTarget Model=Gemini 2.5 Flash2026.010.630.0373
ActorAttackTarget Model=o3 Mini2026.010.630.0312
SirenTarget Model=Qwen 2.5 72B Instruct2026.010.670.0342
ActorAttackTarget Model=GPT-4o2026.010.670.0332
ActorAttackTarget Model=Llama 3.3 70B Instruct2026.010.680.0308
ActorAttackTarget Model=Qwen 2.5 72B Instruct2026.010.680.0318
CrescendoTarget Model=o3 Mini2026.010.680.0292
ActorAttackTarget Model=Average2026.010.680.0326
SirenTarget Model=o3 Mini2026.010.70.0325
SirenTarget Model=GPT-4o2026.010.720.0325
X-TeamingTarget Model=o3 Mini2026.010.720.0368
ActorAttackTarget Model=Gemini 2.5 Flash2026.010.750.036
CrescendoTarget Model=Average2026.010.790.0316
X-TeamingTarget Model=Gemini 2.5 Flash2026.010.80.0373
CrescendoTarget Model=GPT-4o2026.010.80.0308
CrescendoTarget Model=Llama 3.3 70B Instruct2026.010.820.0301
X-TeamingTarget Model=Average2026.010.820.0359
CrescendoTarget Model=Qwen 2.5 72B Instruct2026.010.830.032
CrescendoTarget Model=Gemini 2.5 Flash2026.010.830.0358
X-TeamingTarget Model=Llama 3.3 70B Instruct2026.010.850.033
X-TeamingTarget Model=Qwen 2.5 72B Instruct2026.010.850.0347
MastermindTarget Model=Gemini 2.5 Flash2026.010.880.0393
X-TeamingTarget Model=GPT-4o2026.010.880.0377
MastermindTarget Model=Llama 3.3 70B Instruct2026.010.90.0365
MastermindTarget Model=o3 Mini2026.010.90.0372
MastermindTarget Model=Average2026.010.910.0379
MastermindTarget Model=Qwen 2.5 72B Instruct2026.010.930.038
MastermindTarget Model=GPT-4o2026.010.930.0385