OpenAI MentalHealthBench - User Profile - Clinician: leaderboard
User-profile split of OpenAI's MentalHealthBench (launch post and paper, 2026-09-23): the 70 conversations with a clinician user. Mean task-clipped rubric score, 0-100, higher is better: GPT-5.6 Sol grades each reply against expert-written signed rubrics, and models run at their API default reasoning effort. GPT-5 Thinking and Gemini 2.5 Flash come from the paper's Figure 10.
Metric: Task-clipped rubric score (%). Source: openai.com. Saturation forecast: Around 2030. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6 Astra | 65.15 | #5 |
| 2 | GPT-6 Sol | 58.18 | #17 |
| 3 | GPT-6 Luna | 56.22 | #62 |
| 4 | Grok 4.7 | 56.08 | #26 |
| 5 | GPT-5.6 Sol (August 2026) | 55.7 | |
| 6 | Muse Spark 1.3 | 53.4 | #31 |
| 7 | Claude Opus 5.5 | 52.54 | #3 |
| 8 | GPT-5.6 Luna (August 2026) | 49.45 | |
| 9 | Claude Fable 5.1 | 47.76 | #10 |
| 10 | GPT-5 (Thinking) | 47.7 | #108 (GPT-5) |
| 11 | Gemini 2.5 Pro | 47.41 | #161 |
| 12 | Gemini 3.8 Flash | 46.62 | #30 |
| 13 | Claude Sonnet 5 | 46.26 | #67 |
| 14 | Gemini 2.5 Flash | 45.1 | #255 |
| 15 | Gemini 3.1 Pro (Preview) | 45.09 | #66 |
No result here: #6 GPT-6.1 Sol, #9 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=openai-mentalhealthbench-user-profile-clinician · How It Works · Data refreshed daily, snapshot 2026-11-07.