MentalBench (DSM Diagnosis) - Complete Clinical Summaries: leaderboard
Metric: Accuracy (%) on the 1,725 Type 1 cases (structured professional summaries that meet all DSM-5 criteria; single answer); cases synthesized from a psychiatrist-built DSM-5 knowledge graph by three LLMs and validated by a psychiatrist and a clinical psychologist; four options per case; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 22 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.1 | 94.38 | #131 |
| 2 | Claude Sonnet 4.5 | 94.32 | #138 |
| 3 | Claude Haiku 4.5 | 94.2 | #271 |
| 4 | Qwen 2.5 14B Instruct | 94.15 | #634 |
| 5 | Gemini 2.5 Pro | 93.91 | #145 |
| 6 | Qwen 2.5 72B Instruct | 93.85 | #436 |
| 7 | Gemini 2.5 Flash | 93.39 | #237 |
| 8 | Gemma 3 27B (IT) | 92.87 | #509 |
| 9 | Qwen 2.5 32B Instruct | 92.87 | #491 |
| 10 | Gemma 3 12B (IT) | 92.81 | #655 |
| 11 | Llama 3.1 70B Instruct | 91.94 | #548 |
| 12 | Qwen 3 14B | 90.78 | #524 |
| 13 | Qwen 3 32B | 90.2 | #424 |
| 14 | Qwen 3 8B | 89.74 | #667 |
| 15 | Qwen 2.5 7B Instruct | 85.85 | #846 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=mentalbench-dsm-diagnosis-complete-clinical-summaries · How It Works · Data refreshed daily, snapshot 2026-10-11.