Jailbreak Safety Evaluation on English dataset Text
0.01StrongREJECT RateClaude 3 Haiku
Evaluation Results
| Method | Links | |
|---|---|---|
| Claude 3 Haiku2025.04 | 0.01 | |
| Claude 3.5 Sonnet2025.04 | 0.01 | |
| Gemini 1.5 Flash2025.04 | 0.02 | |
| Gemini 1.5 Pro2025.04 | 0.02 | |
| GPT-4o2025.04 | 0.07 | |
| GPT-4o Mini2025.04 | 0.07 |