List your company

Pairwise Prediction Success Rate on Jailbreak Prompt Dataset

78.95Proxy ASRNarrow Safety Proxy
68.768471.411774.05576.6983Nov 27, 2025

Evaluation results

MethodLinks
Narrow Safety ProxyTarget Model=Llama2025.1178.9590.71
Narrow Safety ProxyTarget Model=Qwen2025.1173.4979.23
Narrow Safety ProxyTarget Model=GPT-4o Mini2025.1169.1691.1