List your company

Jailbreak Defense Evaluation on WGT

2.595Strong-Reject Score (SR)vanilla
1.23781.590151.94252.29485Feb 14, 2026

Evaluation results

MethodLinks
vanillaBase Model=Mistral-7b-U2026.022.595
vanillaBase Model=Llama2-7b-U2026.022.551
SCANSBase Model=Mistral-7b-U2026.022.519
SCANSBase Model=Llama2-7b-U2026.022.441
SAGEBase Model=Llama2-7b-U2026.022.425
SAGEBase Model=Mistral-7b-U2026.022.401
vanillaBase Model=Llama3.1-8b-D2026.022.219
SAGEBase Model=Llama3.1-8b-D2026.022.207
SCANSBase Model=Llama3.1-8b-D2026.022.164
ICDBase Model=Llama2-7b-U2026.022.163
SelfDefenDBase Model=Llama2-7b-U2026.022.158
InstructBase Model=Mistral-7b-U2026.022.148
SelfDefenDBase Model=Llama3.1-8b-D2026.021.993
SelfDefenDBase Model=Mistral-7b-U2026.021.813
ICDBase Model=Mistral-7b-U2026.021.749
ICDBase Model=Llama3.1-8b-D2026.021.744
AISABase Model=Llama2-7b-U2026.021.726
InstructBase Model=Llama3.1-8b-D2026.021.621
AISABase Model=Llama3.1-8b-D2026.021.414
AISABase Model=Mistral-7b-U2026.021.346
InstructBase Model=Llama2-7b-U2026.021.29