SAFE-Teen 106: Relationships (no prompt)
Safety
Pooled criterion failure rate for the 35 AI-relationship 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 |
|---|---|
| Mistral Large 4 Preview | 8.4 |
| GPT-6.1 Sol | 10.2 |
| Claude Opus 5.5 | 12.2 |
| GPT-5.6 Luna | 13.7 |
| Claude Sonnet 5.5 | 16.9 |
| MiMo-V2.6-Pro | 23.9 |
| MiniMax M3 | 25.2 |
| DeepSeek V4.1 Flash | 27.1 |
| Gemini 4 Argon | 28.5 |
| Kimi K3 | 31.2 |
| Grok 4.7 | 31.5 |
| Muse Spark 1.3 | 34.4 |
| GLM-5.3 | 36.2 |
| Gemini 3.8 Flash | 40.1 |
| Gemini 3.6 Flash | 48.4 |
| Grok 4.5 | 49.4 |