List your company

Agent Safety Evaluation on AgentHarm Libra

83ScoreSafeMCP
34.1246.8159.572.19Jun 1, 2026

Evaluation results

MethodLinks
SafeMCPBase Model=GPT-4o2026.068366
SafeMCPBase Model=GPT-4o-mini2026.067855
RL-GuardBase Model=GPT-4o2026.067559
GPT-4oBase Model=GPT-4o2026.067260
Llama Guard 3Base Model=GPT-4o2026.067258
SafironBase Model=GPT-4o2026.067260
GPT-4o-miniBase Model=GPT-4o-mini2026.067252
RL-GuardBase Model=GPT-4o-mini2026.067254
Llama Guard 3Base Model=GPT-4o-mini2026.067054
SafironBase Model=GPT-4o-mini2026.066951
LLaMA-3.1-Instruct-8BBase Model=LLaMA-3.1-Instruct-8B2026.066737
NeMoGuard-8bBase Model=GPT-4o2026.066653
SafironBase Model=LLaMA-3.1-Instruct-8B2026.066638
SafeMCPBase Model=LLaMA-3.1-Instruct-8B2026.066635
NeMoGuard-8bBase Model=GPT-4o-mini2026.066450
RL-GuardBase Model=LLaMA-3.1-Instruct-8B2026.066433
Llama Guard 3Base Model=LLaMA-3.1-Instruct-8B2026.065835
NeMoGuard-8bBase Model=LLaMA-3.1-Instruct-8B2026.065835
AgentMonitorBase Model=GPT-4o2026.065548
AgentMonitorBase Model=GPT-4o-mini2026.065445
Qwen3Guard-8BBase Model=GPT-4o2026.065044
Qwen3Guard-8BBase Model=GPT-4o-mini2026.064943
AgentMonitorBase Model=LLaMA-3.1-Instruct-8B2026.064332
Qwen3Guard-8BBase Model=LLaMA-3.1-Instruct-8B2026.064132
Lakera-ChainGuardBase Model=GPT-4o2026.064034
Lakera-ChainGuardBase Model=GPT-4o-mini2026.063933
Lakera-ChainGuardBase Model=LLaMA-3.1-Instruct-8B2026.063630