OpenAI MentalHealthBench: leaderboard
OpenAI benchmark of 1,215 synthetic mental health conversations, from everyday stress to emergencies, with adult, teen, caregiver and clinician users in 11 languages. GPT-5.6 Sol grades the reply to the last message against rubrics written by licensed psychologists and psychiatrists; score is the mean task-clipped share of positive rubric points.
Metric: Task-clipped rubric score (%). Source: openai.com. Saturation forecast: Around 2029. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6 Astra | 57.33 | #5 |
| 2 | GPT-6 Sol | 53.94 | #17 |
| 3 | Claude Opus 5.5 | 52.38 | #3 |
| 4 | GPT-6 Luna | 50.21 | #62 |
| 5 | Muse Spark 1.3 | 48.6 | #31 |
| 6 | GPT-5.6 Sol (August 2026) | 46.98 | |
| 7 | Claude Fable 5.1 | 46.36 | #10 |
| 8 | GPT-5.6 Luna (August 2026) | 44.89 | |
| 9 | Claude Sonnet 5 | 44.54 | #67 |
| 10 | GPT-5 (Thinking) | 42.9 | #108 (GPT-5) |
| 11 | Claude Haiku 4.5 | 41.73 | #291 |
| 12 | Grok 4.7 | 41.3 | #26 |
| 13 | Gemini 3.8 Flash | 35.5 | #30 |
| 14 | Gemini 2.5 Flash | 33.5 | #255 |
| 15 | GPT-4o (Mar 2025) | 32.08 | #339 |
No result here: #6 GPT-6.1 Sol, #9 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=openai-mentalhealthbench · How It Works · Data refreshed daily, snapshot 2026-11-07.