StatFormBench - Coarse-Grained Classification: leaderboard
Accuracy at identifying the implied statistical problem type among 20 coarse-grained categories from informal scenarios with heterogeneous data, on 1,013 StatFormBench samples (StatFormBench, arXiv 2609.01982v2, Table 1). Percent; higher is better.
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Around February 2027. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 79 | #64 |
| 2 | Claude Opus 4.6 | 77.7 | #67 |
| 3 | Gemini 3 Flash | 77 | #103 |
| 4 | GPT-5.4 | 75.2 | #83 |
| 5 | GPT-5.5 | 72.9 | #33 |
| 6 | Claude Sonnet 4.6 | 71.9 | #94 |
| 7 | Kimi K2.5 | 71.5 | #152 |
| 8 | DeepSeek V4 Flash | 71.2 | #155 |
| 9 | Qwen 3.5 397B A17B | 70.6 | #154 |
| 10 | DeepSeek V4 Pro | 66.2 | #112 |
| 11 | Qwen 3.5 4B | 37.7 | #514 |
| 12 | Qwen 3.5 9B | 29.1 | #391 |
| 13 | Qwen 3.5 0.8B | 7.9 | #1410 |
| 14 | Qwen 3.5 2B | 7.7 | #1114 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=statformbench-coarse-grained-classification · How It Works · Data refreshed daily, snapshot 2026-11-05.