ClosureBench: leaderboard
Metric: Accuracy (%; lenient scoring of one JSON answer per question; all three compositional levels, instance-weighted; held-out split of 1,300 generated graph questions over graph sizes 4-10, temperature 0; direct mode: answer immediately). Source: arxiv.org. Saturation forecast: Estimated already saturated. 9 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | O3 | 92.8 |
| 2 | Gemini 2.5 Flash | 87.4 |
| 3 | Claude Sonnet 4 | 81.7 |
| 4 | GPT-4.1 | 52.4 |
| 5 | GPT-4.1 Mini | 49.2 |
| 6 | Qwen 3.5 9B | 43.3 |
| 7 | Qwen 3 4B | 39.5 |
| 8 | Qwen 3.5 2B | 33 |
| 9 | Qwen 2.5 1.5B Instruct | 25.8 |
Interactive version: theaggregate.ai/benchmark?slug=closurebench · How It Works · Data refreshed daily, snapshot 2026-09-26.