MindBench - Alignment: leaderboard
MindBench.ai: models rate 500 fixed chatbot replies from mental-health conversations on a -3 to +3 appropriateness scale, compared with a consensus of 167 clinicians and people with lived experience. Alignment is 1 - MAE/3 against that consensus, the site's headline; 0-1, higher is better. It measures judging replies, not writing them.
Metric: Alignment with human consensus (0-1, 1 - MAE / 3, higher is better). Source: mindbench.ai. Saturation forecast: Not forecast. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O3 | 0.83 | #132 |
| 2 | GPT-5.5 | 0.75 | #33 |
| 3 | Grok 4.3 | 0.74 | #204 |
| 4 | Claude Haiku 4.5 (20251001) | 0.73 | #264 |
| 5 | GPT-5.4 | 0.72 | #83 |
| 6 | Mistral Medium 3.5 | 0.72 | #375 |
| 7 | DeepSeek V4 Flash | 0.7 | #155 |
| 8 | Sonar | 0.7 | #546 |
| 9 | DeepSeek V4 Pro | 0.69 | #112 |
| 10 | Claude Fable 5 | 0.68 | #20 |
| 11 | Claude Opus 4.7 | 0.67 | #51 |
| 12 | Claude Sonnet 4.6 | 0.66 | #94 |
| 13 | Gemini 3.5 Flash | 0.66 | #66 |
| 14 | Mistral Small 4 | 0.66 | #522 |
| 15 | Grok 4.1 Fast | 0.65 | #227 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mindbench-alignment · How It Works · Data refreshed daily, snapshot 2026-11-05.