K-Bench - Substance Misuse: leaderboard
The model talks with a simulated patient presenting substance misuse for 20 turns under the providers' Default v1 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. 60 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 (High) | 97.6 | #49 (GPT-5.5) |
| 2 | GPT-5.5 (Medium) | 96.97 | #49 (GPT-5.5) |
| 3 | GPT-5.2 (Non-reasoning) | 96.73 | #176 (GPT-5.2) |
| 4 | GPT-5.5 (Low) | 96.52 | #49 (GPT-5.5) |
| 5 | Claude Fable 5 (High) | 96.19 | #27 (Claude Fable 5) |
| 6 | GPT-5.4 (Low) | 95.6 | #123 (GPT-5.4) |
| 7 | GPT-5.2 (High) | 95.58 | #176 (GPT-5.2) |
| 8 | GPT-5.5 (Non-reasoning) | 95.35 | #49 (GPT-5.5) |
| 9 | GPT-5.6 Terra (Non-reasoning) | 95.16 | #62 (GPT-5.6 Terra) |
| 10 | GPT-5.6 Terra (Medium) | 95 | #62 (GPT-5.6 Terra) |
| 11 | GPT-5.4 (High) | 94.89 | #123 (GPT-5.4) |
| 12 | GPT-5.6 Luna (Non-reasoning) | 94.55 | #121 (GPT-5.6 Luna) |
| 13 | GPT-5.6 Sol (Non-reasoning) | 94.17 | #30 (GPT-5.6 Sol) |
| 14 | GPT-5.6 Sol (Medium) | 94.11 | #30 (GPT-5.6 Sol) |
| 15 | GPT-5.4 (Non-reasoning) | 93.99 | #123 (GPT-5.4) |
Interactive version: theaggregate.ai/benchmark?slug=k-bench-substance-misuse · How It Works · Data refreshed daily, snapshot 2026-11-13.