PureDocBench - Clean - Formula CDM: leaderboard
Metric: Formula Character Detection Matching score (0-100) between the rendered predicted and reference LaTeX formulas, on the clean track (pages rendered straight from their HTML/CSS source) of 1,475 source pages in 10 domains and 66 subcategories, rendered from LLM-written HTML/CSS with source-linked annotations for text, formulas, tables and reading order; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 58 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GLM-5.3 Flash | 80.16 |
| 2 | Gemini 3.6 Flash | 78.29 |
| 3 | Claude Opus 5 | 75.92 |
| 4 | GPT-5.6 Sol | 74.96 |
| 5 | Qwen3.8-Flash-Next | 72.28 |
| 6 | Qwen 3.8 27B | 71.68 |
| 7 | Qwen 3.6 35B A3B | 70.39 |
| 8 | Qwen 3.5 4B | 69.96 |
| 9 | Qwen 3.6 27B | 69.36 |
| 10 | Qwen 3.5 122B A10B | 67.96 |
| 11 | Qwen 3.5 9B | 67.6 |
| 12 | Kimi K2.6 (Non-reasoning) | 66.93 |
| 13 | Qwen 3.5 27B | 66.36 |
| 14 | Gemini 3.1 Pro (Preview) | 65.63 |
| 15 | Qwen 3.5 397B A17B | 65.26 |
Interactive version: theaggregate.ai/benchmark?slug=puredocbench-clean-formula-cdm · How It Works · Data refreshed daily, snapshot 2026-10-07.