BehaviorBench - Economics Olympiad: leaderboard
BehaviorBench (University of Michigan, 2026) evaluates foundation models on behavioral science tasks at the individual and population levels, using economic-game records, Big Five personality surveys and research literature. This board is accuracy (%) on 124 four-option International Economics Olympiad problems from 2018 to 2025, averaged over five runs; higher is better.
Metric: Accuracy (%) on 124 four-option multiple-choice International Economics Olympiad problems from 2018 to 2025, mean of five runs; chance 25; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 95.6 | #94 |
| 2 | Claude Opus 4.6 | 95.6 | #102 |
| 3 | GPT-5.4 (High) | 95.6 | #123 (GPT-5.4) |
| 4 | GPT-5.4 Mini (High) | 93.2 | #389 (GPT-5.4 Mini) |
| 5 | Claude Sonnet 4.6 | 83.1 | #161 |
| 6 | GPT-4.1 | 78.4 | #495 |
| 7 | Gemini 3.1 Flash Lite (Preview) | 77.6 | #418 |
| 8 | DeepSeek V3.2 | 74.4 | #406 |
| 9 | Qwen 3 4B | 73.9 | #2281 |
| 10 | Claude Haiku 4.5 | 70 | #568 |
| 11 | Llama 3.3 70B Instruct | 65.8 | #1357 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=behaviorbench-economics-olympiad · How It Works · Data refreshed daily, snapshot 2026-11-13.