List your company

Red-teaming on AdvBench (test)

88ASRAMIS
0.6423.324668.68Nov 3, 2025

Evaluation results

MethodLinks
AMISAttacker Model=Llama-3.1-8B, Target Model=Claude-3.5-Haiku2025.118842
ActorBreakerAttacker Model=GPT-4o-mini, Target Model=Claude-3.5-Haiku2025.114823
Rainbow TeamingIterations=500, Target Model=Claude-3.5-Haiku2025.11383
ActorBreakerAttacker Model=Llama-3.1-8B, Target Model=Claude-3.5-Haiku2025.113618
Rainbow TeamingIterations=300, Target Model=Claude-3.5-Haiku2025.11342
Rainbow TeamingIterations=100, Target Model=Claude-3.5-Haiku2025.11100
ActorBreakerAttacker Model=Llama-3.1-70B, Target Model=Claude-3.5-Haiku2025.1162
ActorBreakerAttacker Model=Claude-3.5-Haiku, Target Model=Claude-3.5-Haiku2025.1144