Benchmark Results

Results Summary

Tool	NID	TEDS	MHS	Overall	Speed
EdgeParse	0.911	0.783	0.818	0.880	0.026s
Docling	0.899	0.887	0.824	0.882	1.274s
Marker	0.866	0.825	0.794	0.846	30.34s
EdgeQuake	0.878	0.795	0.685	0.828	6.725s
OpenDataLoader	0.912	0.494	0.760	0.844	0.053s
PyMuPDF4LLM	0.888	0.540	0.774	0.833	0.723s
MarkItDown	0.844	0.273	0.000	0.589	0.197s

Key Takeaways

• EdgeParse is the fastest — 0.026s per document (49× faster than Docling)
• Highest overall among rule-based tools — 0.880 without any ML model
• Competitive with ML tools — within 0.2% of Docling’s overall score
• Best NID score — 0.911, matching OpenDataLoader for reading order accuracy
• Best rule-based TEDS — 0.783 for table structure

Speed Comparison

Comparison	Factor
EdgeParse vs Docling	49× faster
EdgeParse vs PyMuPDF4LLM	28× faster
EdgeParse vs Marker	1,167× faster
EdgeParse vs EdgeQuake	259× faster

Test Environment

• Hardware: Apple M4 Max
• Corpus: 200 diverse PDF documents
• Mode: Single-threaded
• Categories: Academic papers, financial reports, invoices, government forms, books, manuals

See It in Action

• Try Live Demo — parse any PDF in your browser with WebAssembly, no install needed
• Enterprise — production-grade deployment with priority support by Elitizon