List your company

Jailbreak Attack on Llama 8B 3.1

96NR RateSWhisper
-3.8422.084873.92Mar 14, 2026

Evaluation results

MethodLinks
SWhisperSurrogate Model=LLaMA-3.1-8B2026.039694.5
SWhisperSurrogate Model=Qwen2.5-7B2026.035857.8
SWhisperSurrogate Model=Gemma-3-4B2026.035452.5
SWhisperSurrogate Model=Mistral-7B2026.034645.8
Auto-DANSurrogate Model=LLaMA-3.1-8B2026.033029.8
Auto-DANSurrogate Model=Mistral-7B2026.032827
Auto-DANSurrogate Model=Qwen2.5-7B2026.031818
PAIRSurrogate Model=Mistral-7B2026.031613.5
Auto-DANSurrogate Model=Gemma-3-4B2026.031616
PAIRSurrogate Model=LLaMA-3.1-8B2026.031410.5
PAIRSurrogate Model=Gemma-3-4B2026.031210.5
PAIRSurrogate Model=Qwen2.5-7B2026.03109.5
GCGSurrogate Model=LLaMA-3.1-8B2026.0322
GCGSurrogate Model=Mistral-7B2026.0322
IGCGSurrogate Model=Qwen2.5-7B2026.0322
GCGSurrogate Model=Gemma-3-4B2026.0300
GCGSurrogate Model=Qwen2.5-7B2026.0300
IGCGSurrogate Model=LLaMA-3.1-8B2026.0300
IGCGSurrogate Model=Gemma-3-4B2026.0300
IGCGSurrogate Model=Mistral-7B2026.0300