List your company

Jailbreak Defense on X-Teaming

0Attack Success Rate (ASR)Prompt Guard
-3.6420.9345.570.07Apr 1, 2026

Evaluation results

MethodLinks
Prompt GuardLatency (Sec.)=22.67, Memory Overhead (MB)=1864.112026.0400
SelfGraderLatency (Sec.)=1.24, Memory Overhead (MB)=479.812026.0400
GradientCuffLatency (Sec.)=26.57, Memory Overhead (MB)=2063.762026.0411
SelfDefend (Intent)Latency (Sec.)=3.11, Memory Overhead (MB)=13457.402026.042222
SelfDefend (Direct)Latency (Sec.)=1.19, Memory Overhead (MB)=13452.932026.045961
GuardReasoner (Pre)Latency (Sec.)=14.87, Memory Overhead (MB)=15617.332026.046465
GuardReasoner (Post)Latency (Sec.)=16.71, Memory Overhead (MB)=15618.132026.047683
Llama Guard (Pre)Latency (Sec.)=0.96, Memory Overhead (MB)=13910.612026.047781
Llama Guard (Post)Latency (Sec.)=1.17, Memory Overhead (MB)=13910.612026.048591
LLama-3-8B-Instruct (No Defense)Latency (Sec.)=1.472026.0491—
Perplexity FilterLatency (Sec.)=0.49, Memory Overhead (MB)=13571.942026.0491100