Dr. DocBench - Table TEDS: leaderboard
Dr. DocBench (2026) parses 4,514 difficult pages sampled by parser failure from long books in 52 subject domains, with expert annotations of text, formulas, tables and reading order. This board is the tree-edit-distance similarity (TEDS) of tables rendered as HTML, content and structure, on a 0 to 100 scale; higher is better.
Metric: Tree-edit-distance similarity of tables converted to HTML, content and structure (0-100), 4,514 expert-annotated difficult pages sampled by parser failure from long books in 52 subject domains; each page parsed with the model's own document-to-markdown prompt (specialized parsers through their official APIs); higher is better. Source: arxiv.org. Saturation forecast: Around September 2027. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Kimi K2.5 | 51.98 | #256 |
| 2 | Gemini 3.1 Pro (Preview) | 51.26 | #94 |
| 3 | Qwen 3.5 Plus | 49.77 | #232 |
| 4 | Claude Opus 4.6 | 49.21 | #102 |
| 5 | GPT-5.5 | 48.9 | #49 |
| 6 | Qwen 3.5 Flash | 46.24 | #435 |
| 7 | Qwen 3.5 122B A10B | 38.31 | #331 |
| 8 | GPT-4o | 30.14 | #741 |
| 9 | Nemotron Nano 12B v2 VL | 27.03 | #2231 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=dr-docbench-table-teds · How It Works · Data refreshed daily, snapshot 2026-11-13.