List your company

Jailbreak Attack on Safety Evaluation Benchmark Harmful Categories

0ASR (IA)FigStep
-2.7215.643452.36Dec 5, 2025

Evaluation results

MethodLinks
FigStepModel=Claude-4.5-sonnet2025.1201.281.62121.581.9261.28161.88342.48121.7
MM-safetybenchModel=Claude-4.5-sonnet2025.120101.1201.9141.641.2121.72161.79.431.47
MM-safetybenchModel=GPT-4.12025.1221.12242.1382.6422.7322.3543.34262.1231.142.33
MM-safetybenchModel=Gemini-2.5-pro2025.1221.1161.8362.6282.2101.6483.34101.5221.432.01
FigStepModel=Gemini-2.5-pro2025.1241.2441.4821.121.18161.72362.92121.6410.861.62
FigStepModel=GPT-4o2025.1261.5241.520101.1841.56443.34161.8810.571.71
FigStepModel=GPT-4.12025.1281.5821.44101.701.361.54483.3161.9212.861.86
HADESModel=Claude-4.5-sonnet2025.12121.8101.68242.2161.821.221.220212.291.68
MM-safetybenchModel=GPT-4o2025.12201.26202.14402.77503.08362.62563.68221.9834.862.5
HADESModel=Gemini-2.5-pro2025.12242.08242.14181.9201.98262.5221.96262.122.862.1
HADESModel=GPT-4.12025.12362.6221.96281.9342.7282.54282.34282.1429.142.31
PRISMModel=Claude-4.5-sonnet2025.12383.1403.1463.5503.5383.2282.8523.841.713.32
PRISMModel=GPT-4o2025.12423.48383.2423.56604.14323.18222.78523.6441.143.43
HADESModel=GPT-4o2025.12503.2182.04403.1222.1242.28242.5282.2829.432.5
VRSAModel=Claude-4.5-sonnet2025.12523.62463.3583.7563.52523.48423.28603.8452.293.53
PRISMModel=GPT-4.12025.12604443.6563.9644483.4323.2563.851.433.74
VRSAModel=Gemini-2.5-pro2025.12643.74543.8583.74543.5603.72623.92683.9603.76
VRSAModel=GPT-4o2025.12663.94603.86563.6623.76563.76644.06643.7261.143.81
PRISMModel=Gemini-2.5-pro2025.12663.9443.5543.8683.9543.7182.5523.450.863.53
VRSAModel=GPT-4.12025.12684.1523.58684623.8583.54623.94683.9662.573.82