InduOCRBench - Table: leaderboard

The table recognition score of InduOCRBench (2026), a document-parsing benchmark of 570 real industrial documents in 11 challenge types: tree-edit-distance similarity (TEDS) between predicted and reference tables, 0 to 100. Higher is better.

Metric: Table recognition score (0-100): Tree-Edit-Distance-based Similarity (TEDS) between the predicted and reference HTML tables on 570 industrial documents (3,402 pages) in 11 challenge types (complex background, high pixel count, ultra-long, multi-column, ultra-wide, historical books, handwriting, multi-font, visual style, watermark, cross-page tables) plus normal documents, each OCR method converting pages to structured Markdown (general VLMs with a standardized prompt, pipeline tools at default configuration, closed solutions through their official APIs); higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 10 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1PaddleOCR-VL72
2Gemini 2.5 Pro63.3#276
3MinerU2.5-2509-1.2B60.3
4Mineru2-pipeline56.3
5Qwen 3 VL 235B A22B Instruct54.6#532
6HunyuanOCR52.5
7Doc2X52.1
8PP-StructureV349.1
9DeepSeek-OCR47.1
10GPT-4o37.2#751

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

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