OpenAI MentalHealthBench - Behavior - Urgency Calibration: leaderboard
Behavior axis of OpenAI's MentalHealthBench (launch post and paper, 2026-09-23): matching concern, escalation and referral to the severity of the situation, avoiding both over-alarm and under-response, scored on the 285 rubric criteria tagged with this axis. The post's axis score is the signed rubric score on this axis, normalized to its possible range (0-100, higher is better); GPT-5.6 Sol grades, models run at API default reasoning effort. The post's 15 models only: the paper's behavior figure uses a different scale.
Metric: Axis score, normalized to its possible range (%). Source: openai.com. Saturation forecast: Around October 2028. 15 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.8 Flash | 81.44 | #30 |
| 2 | GPT-6 Sol | 80.53 | #17 |
| 3 | Claude Fable 5.1 | 78.08 | #10 |
| 4 | GPT-6 Astra | 77.66 | #5 |
| 5 | GPT-5.6 Sol (August 2026) | 77.25 | |
| 6 | GPT-6 Luna | 76.77 | #62 |
| 7 | GPT-5.6 Luna (August 2026) | 75.7 | |
| 8 | Gemini 3.1 Pro (Preview) | 75.43 | #66 |
| 9 | Claude Sonnet 5 | 75.38 | #67 |
| 10 | Muse Spark 1.3 | 74.85 | #31 |
| 11 | Claude Opus 5.5 | 73.97 | #3 |
| 12 | Gemini 2.5 Pro | 73.86 | #161 |
| 13 | Grok 4.7 | 71.72 | #26 |
| 14 | Claude Haiku 4.5 | 69.92 | #291 |
| 15 | GPT-4o (Mar 2025) | 69.65 | #339 |
No result here: #6 GPT-6.1 Sol, #9 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=openai-mentalhealthbench-behavior-urgency-calibration · How It Works · Data refreshed daily, snapshot 2026-11-07.