OpenExempt: leaderboard
Diagnostic legal benchmark for reasoning over U.S. Bankruptcy Code exemption scenarios.
Metric: Mean F1 (self-reported). Source: benchmarklist.com. Status: saturated. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | O3 | 72.31 |
| 2 | Gemini 2.5 Pro | 71.71 |
| 3 | GPT-5 | 71.67 |
| 4 | Gemini 2.5 Flash | 65.76 |
| 5 | DeepSeek R1 | 64.63 |
| 6 | O4 Mini | 59.35 |
| 7 | Claude Sonnet 4 | 57.37 |
| 8 | GPT-4.1 | 47.77 |
| 9 | DeepSeek V3 | 47.67 |
| 10 | Llama 4 Maverick | 40.92 |
| 11 | Claude 3.5 Haiku | 36.13 |
| 12 | Llama 4 Scout | 31.91 |
Interactive version: theaggregate.ai/benchmark?slug=openexempt · How It Works · Data refreshed daily, snapshot 2026-09-05.