List your company

Jailbreak Attack Evaluation on Do-Not-Answer

2.5ASRGemini 2.5 Flash
2.045.1458.2511.355Jan 8, 2026

Evaluation results

MethodLinks
Gemini 2.5 FlashModel=Gemini 2.5 Flash2026.012.5
GPT-4o-miniModel=GPT-4o-mini2026.013
Qwen2.5 7BModel=Qwen2.5 7B2026.013
Grok 3 MiniModel=Grok 3 Mini2026.014
DeepSeek V3.1Model=DeepSeek V3.12026.014.5
Mixtral 8×7BModel=Mixtral 8×7B2026.0114