BehaviorBench - Age from Personality: leaderboard
Metric: Mean absolute error in years of the age of the subject predicted from the five Big Five dimension scores (Big Five personality dimension scores of held-out survey subjects); lower is better. Source: arxiv.org. Saturation forecast: Around July 2027. 18 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama 3.3 70B Instruct | 9.3 |
| 2 | Gemini 3.1 Pro (Preview) | 10.5 |
| 3 | GPT-5.4 (High) | 10.5 |
| 4 | GPT-4.1 | 10.6 |
| 5 | DeepSeek V3.2 | 10.7 |
| 6 | GPT-5.4 Mini (High) | 10.9 |
| 7 | Qwen 3 4B | 11.2 |
| 8 | Claude Haiku 4.5 | 11.8 |
| 9 | Claude Opus 4.6 | 13.3 |
| 10 | Gemini 3.1 Flash Lite (Preview) | 14 |
| 11 | Claude Sonnet 4.6 | 14.5 |
Interactive version: theaggregate.ai/benchmark?slug=behaviorbench-age-from-personality · How It Works · Data refreshed daily, snapshot 2026-09-29.