K-Bench - Suicide Risk (Therapeutic v0): leaderboard
The model talks with a simulated patient presenting suicide risk for 20 turns under the Therapeutic v0 system prompt; the number is the 0-100 risk-handling score (clinical judgement and risk exploration) on that domain's vignettes. Unlike the K-Bench - Risk board, which pools the four risk domains, it shows the domain alone.
Metric: Risk-handling score in this domain (0-100, higher is better). Source: k-bench.ai. 61 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 (Non-reasoning) | 96.36 | #176 (GPT-5.2) |
| 2 | GPT-5.4 (Low) | 95.89 | #123 (GPT-5.4) |
| 3 | GPT-5.5 (Medium) | 95.2 | #49 (GPT-5.5) |
| 4 | Claude Fable 5 (Low) | 95.2 | #27 (Claude Fable 5) |
| 5 | GPT-5.4 (Non-reasoning) | 95 | #123 (GPT-5.4) |
| 6 | Claude Fable 5 (High) | 94.73 | #27 (Claude Fable 5) |
| 7 | GPT-5.2 (High) | 94.55 | #176 (GPT-5.2) |
| 8 | Qwen 3 32B | 94.14 | #960 |
| 9 | Claude Haiku 4.5 | 94.06 | #568 |
| 10 | Claude Opus 4.8 (Low) | 93.93 | #56 (Claude Opus 4.8) |
| 11 | GPT-5.5 (High) | 93.54 | #49 (GPT-5.5) |
| 12 | GPT-5.4 (High) | 93.51 | #123 (GPT-5.4) |
| 13 | Gemini 3.5 Flash (High) | 93.51 | #101 (Gemini 3.5 Flash) |
| 14 | GPT-5.5 (Low) | 93.36 | #49 (GPT-5.5) |
| 15 | Claude Opus 4.8 (High) | 93.26 | #56 (Claude Opus 4.8) |
Interactive version: theaggregate.ai/benchmark?slug=k-bench-suicide-risk-therapeutic-v0 · How It Works · Data refreshed daily, snapshot 2026-11-13.