PureDocBench - Real Degraded - Table TEDS: leaderboard

Table recognition on the real degraded track of PureDocBench (2026), a document-parsing benchmark of 1,475 source-traceable pages (pages captured by phone from prints, photocopies and screens, or as compressed screenshots), scored by tree-edit-distance similarity (TEDS) from 0 to 100. Higher is better.

Metric: Table Tree-Edit-Distance-based Similarity (0-100) between the predicted and reference HTML tables, on the real degraded track (the same pages through four physical acquisition chains: phone photos of prints and of photocopies, screen photographs, and screenshots with social-media compression) 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 May 2027. 58 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Claude Opus 584.01#26
2GLM-5.3 Flash82.45#133
3Gemini 3.6 Flash81.96#81
4Qwen 3.8 27B79.63#221
5WeVisDoc-4B77.92
6Gemini 3.1 Pro (Preview)77.26#99
7TeleOCR76.32
8Qwen 3.5 122B A10B75.44#346
9Kimi K2.6Non-reasoning75.14#192
10WeVisDoc-2B74.21
11OvisOCR273.7
12MinerU-2.5-Pro-1.2B72.47
13FD-RL72.08
14PaddleOCR-VL-1.671.76
15Logics-Parsing-v271.64

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=puredocbench-real-degraded-table-teds · How It Works · Data refreshed daily, snapshot 2026-11-21.