Skip to content

Benchmark Results

EdgeParse publishes two scoreboards. They use different overall formulas — do not mix numbers across boards.

BoardExportFormulaRole
EdgeParse harnessbenchmarkSnapshotProduct overall (quality + structure)Regression / product score
Official OpenDataLoader board (odl-bench)odlBenchSnapshotmean(NID, TEDS, MHS)Match public OpenDataLoader leaderboard

Updated 2026-10-02 · Apple M4 Max · 200 docs · overall 0.760 · 0.284 s/doc

ToolNIDTEDSMHSOverallSpeed
EdgeParse0.8800.5620.4780.7600.284 s/doc
EdgeParse [hybrid]0.8630.6030.5060.7670.810 s/doc
Docling (IBM)0.8750.5680.4500.7582.332 s/doc
OpenDataLoader0.8730.3200.4410.7330.022 s/doc
PyMuPDF4LLM0.8600.5090.4110.7320.640 s/doc
OpenDataLoader [hybrid]0.8690.4220.4110.7312.537 s/doc
MarkItDown0.8070.1930.0010.5640.189 s/doc
  • Deterministic EdgeParse leads this board at overall 0.760 and 0.284 s/doc
  • 8× faster than Docling · 2× faster than PyMuPDF4LLM · 0.1× faster than OpenDataLoader
  • No ML stack required for born-digital PDFs; optional in-browser OCR for image tables

Updated 2026-10-01 · Apple M4 Max · 200 docs · formula {claims.odl.formula}

EdgeParse [hybrid] leads NID, TEDS, MHS, and overall (0.900).

ToolNIDTEDSMHSOverallStatus
EdgeParse [hybrid]0.9170.9280.8360.900finished
Docling0.9050.9250.8270.891finished
OpenDataLoader [hybrid]0.9150.6760.8220.877finished
PyMuPDF4LLM0.9160.8030.7720.873finished
EdgeParse0.8940.7730.7630.857finished
EdgeParse [wasm]0.8910.7760.7610.854finished
OpenDataLoader0.9120.4830.7570.842finished
MarkItDown0.8440.2730.0000.589finished
Unstructured0.0930.0000.0540.086partial
  • EdgeParse [hybrid] — overall 0.900 (leads every metric)
  • EdgeParse (deterministic) — overall 0.857
  • EdgeParse [wasm] — overall 0.854 (TEDS 0.776)
  • Unstructured is partial (not a finished competitor); Marker / MinerU / Nutrient were not run
ComparisonFactor
EdgeParse vs Docling8.2× faster
EdgeParse vs PyMuPDF4LLM2.3× faster
EdgeParse vs OpenDataLoader0.1× faster
EdgeParse vs MarkItDown0.7× faster
  • Hardware: Apple M4 Max
  • Corpus: 200 diverse PDF documents
  • Harness path: benchmark/
  • Official board path: odl-bench/
  • Categories: Academic papers, financial reports, invoices, government forms, books, manuals
  • Try Live Demo — parse any PDF in your browser with WebAssembly, no install needed
  • Enterprise — production-grade deployment with priority support by Elitizon