Jailbreak Attack Evaluation on StealthGraph SG-Implicit
91ASRGrok 3 Mini
Evaluation Results
| Method | Links | |
|---|---|---|
| Grok 3 MiniModel=Grok 3 Mini2026.01 | 91 | |
| Mixtral 8×7BModel=Mixtral 8×7B2026.01 | 90.5 | |
| Qwen2.5 7BModel=Qwen2.5 7B2026.01 | 88 | |
| GPT-4o-miniModel=GPT-4o-mini2026.01 | 87.5 | |
| DeepSeek V3.1Model=DeepSeek V3.12026.01 | 77.5 | |
| Gemini 2.5 FlashModel=Gemini 2.5 Flash2026.01 | 75 | |
| Mixtral 8×7BModel=Mixtral 8×7B2026.01 | 0.76 | |
| Grok 3 MiniModel=Grok 3 Mini2026.01 | 0.745 | |
| Qwen2.5 7BModel=Qwen2.5 7B2026.01 | 0.635 | |
| GPT-4o-miniModel=GPT-4o-mini2026.01 | 0.57 | |
| DeepSeek V3.1Model=DeepSeek V3.12026.01 | 0.515 | |
| Gemini 2.5 FlashModel=Gemini 2.5 Flash2026.01 | 0.47 |