Self-Harm on Human-Authored
1.93Non-safe RateGPT o4-mini
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT o4-mini2026.05 | 1.93 | |
| Gemini 2.5 flash2026.05 | 4.76 | |
| Claude 4.5 Haiku2026.05 | 8.93 | |
| Llama 4 Scout2026.05 | 15.92 |
| Method | Links | |
|---|---|---|
| GPT o4-mini2026.05 | 1.93 | |
| Gemini 2.5 flash2026.05 | 4.76 | |
| Claude 4.5 Haiku2026.05 | 8.93 | |
| Llama 4 Scout2026.05 | 15.92 |