List your company

Safety Evaluation on PS-Bench base setting (test)

18ASR (Hate Speech)Stateless
16.92424.18731.4538.713Jan 25, 2026

Evaluation results

MethodLinks
StatelessBase Model=Qwen3-8B2026.01182246413208.63
StatelessBase Model=Qwen3-235B-A22B2026.0119458154271211.75
Mem0Base Model=Qwen3-8B2026.0120.35.82.58.410.511.222.330.613.95
LDAgentBase Model=Qwen3-8B2026.0121.786.28.66.114.327.929.415.28
StatelessBase Model=DeepSeek-V3.22026.01231247211229
MemOSBase Model=Qwen3-8B2026.0123.7117.412.211.315.328.938.418.53
StatelessBase Model=GPT-4o2026.0125585301326414.5
MemUBase Model=Qwen3-8B2026.0125.2127.915.116.21628.843.220.55
AmemBase Model=Qwen3-235B-A22B2026.0128.110.112.516.229.918.732.918.620.88
Mem0Base Model=DeepSeek-V3.22026.0129.220.411.116.424.219.335.515.221.41
AmemBase Model=Qwen3-8B2026.0129.99.710.617.514.218.936.640.422.23
LDAgentBase Model=Qwen3-235B-A22B2026.0130.213.514.313.931.221.233.925.823
Mem0Base Model=Qwen3-235B-A22B2026.0135.911.7111529.317.832.65.819.89
AmemBase Model=DeepSeek-V3.22026.0135.939.817.422.233.92836.115.628.61
StatelessBase Model=GPT-4o-mini2026.013616116522234823.13
MemOSBase Model=DeepSeek-V3.22026.0136.429.713.62324.423.641.917.226.23
MemUBase Model=DeepSeek-V3.22026.0136.627.217.127.829.223.343.824.228.65
AmemBase Model=GPT-4o2026.0138.11115.420.140.621.342.27.424.51
Mem0Base Model=GPT-4o-mini2026.0138.217.517.314.252.827.341.45.626.79
MemOSBase Model=Qwen3-235B-A22B2026.0138.319.912.316.631.722.836.47.423.18
MemOSBase Model=GPT-4o-mini2026.0140.321.818.417.847.23145.77.628.73
MemUBase Model=Qwen3-235B-A22B2026.0140.616.814.520.238.721.3389.224.91
MemOSBase Model=GPT-4o2026.0140.813.51926.141.930.547.65.228.08
LDAgentBase Model=DeepSeek-V3.22026.0140.930.819.92337.628.146.12130.93
Mem0Base Model=GPT-4o2026.014114.320.123.739.423.944.49.827.08
LDAgentBase Model=GPT-4o2026.0141.817.219.123.241.430.946.68.228.55
MemUBase Model=GPT-4o2026.0141.914.71825.345.330.350.78.429.33
MemUBase Model=GPT-4o-mini2026.0142.216.619.61851.629.243.37.228.46
LDAgentBase Model=GPT-4o-mini2026.0144.220.317.916.948.527.642.26.828.05
AmemBase Model=GPT-4o-mini2026.0144.919.419.119.350.330.444.59.429.66