Jailbreak Attack on Jailbreak Evaluation GPT-4o-mini
93.33ASRMAPA
Evaluation Results
| Method | Links | |
|---|---|---|
| MAPAVictim Model=GPT-4o-mini, Judge LLM=Advanced Judge (GPT-4o-mini)2026.02 | 93.33 | |
| HPM2025.12 | 89.5 | |
| MAPAVictim Model=GPT-4o-mini, Judge LLM=Default Judge (Llama-2-13B)2026.02 | 88.33 | |
| PAP2025.12 | 86.9 | |
| CoAVictim Model=GPT-4o-mini, Judge LLM=Advanced Judge (GPT-4o-mini)2026.02 | 73.33 | |
| CoA2025.12 | 67.4 | |
| ActorAttackVictim Model=GPT-4o-mini, Judge LLM=Advanced Judge (GPT-4o-mini)2026.02 | 64 | |
| FootInTheDoorVictim Model=GPT-4o-mini, Judge LLM=Advanced Judge (GPT-4o-mini)2026.02 | 60 | |
| PAIR2025.12 | 53.5 | |
| CoAVictim Model=GPT-4o-mini, Judge LLM=Default Judge (Llama-2-13B)2026.02 | 53.33 | |
| ActorAttackVictim Model=GPT-4o-mini, Judge LLM=Default Judge (Llama-2-13B)2026.02 | 45.76 | |
| FootInTheDoorVictim Model=GPT-4o-mini, Judge LLM=Default Judge (Llama-2-13B)2026.02 | 41.66 | |
| AutoDAN2025.12 | 38.5 |