SAFE-Teen 106: Urgent Safety (no prompt)
Safety
Pooled criterion failure rate for the 35 urgent-safety scenarios in the revised 106-scenario cohort, without a teen-aware prompt. Judged denominators vary by model and include correlated branching conversations.
Top models (lower is better)
| Model | Score |
|---|---|
| GPT-6.1 Sol | 3.4 |
| GPT-5.6 Luna | 8.5 |
| Muse Spark 1.3 | 9.0 |
| DeepSeek V4.1 Flash | 11.3 |
| Claude Opus 5.5 | 12.7 |
| Claude Sonnet 5.5 | 14.7 |
| Grok 4.7 | 15.1 |
| Mistral Large 4 Preview | 18.8 |
| MiMo-V2.6-Pro | 19.7 |
| Kimi K3 | 23.0 |
| Grok 4.5 | 26.1 |
| MiniMax M3 | 27.2 |
| GLM-5.3 | 29.3 |
| Gemini 3.8 Flash | 31.5 |
| Gemini 4 Argon | 33.1 |
| Gemini 3.6 Flash | 34.4 |