BehaviorBench - Beauty Contest Win Rate: leaderboard
Metric: Win rate (%) in the beauty contest game played against recorded human players: share of sessions in which the guess of the model is closer to two thirds of the group average than every human guess; higher is better. Source: arxiv.org. Saturation forecast: Around 2038. 18 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6 | 17 |
| 2 | GPT-4.1 | 17 |
| 3 | Claude Opus 4.6 | 15 |
| 4 | DeepSeek V3.2 | 14 |
| 5 | Gemini 3.1 Flash Lite (Preview) | 14 |
| 6 | Claude Haiku 4.5 | 13 |
| 7 | Gemini 3.1 Pro (Preview) | 12 |
| 8 | GPT-5.4 Mini (High) | 10 |
| 9 | Qwen 3 4B | 9 |
| 10 | GPT-5.4 (High) | 8 |
| 11 | Llama 3.3 70B Instruct | 6 |
Interactive version: theaggregate.ai/benchmark?slug=behaviorbench-beauty-contest-win-rate · How It Works · Data refreshed daily, snapshot 2026-09-29.