📃 Extensive List of Document Parsers

• 🚧 THIS IS A WORK IN PROGRESS! More will be added soon!
• Feel free to contribute by submitting a pull request 🙏
• Cells marked with ✅ or ❌ have been independently tested. Blank cells indicate that the feature has not yet been independently tested.
• See the results folder to see the outputs from models.

PDF-to-Text Converters

Usually outputs as raw text or markdown

Machine-generated Documents only

Models	Source	Output	Needs prompt?	Table	Equation	Figure	Handwriting	Two columns	Multiple columns
PyMuPDF		Raw text	N	❌	❌	❌	❌	✅	❌
PDFPlumber		Raw text	N	✅ (separate from text)	❌	❌	❌	❌	❌

Machine-generated and Scanned Documents

Models	Source	Output	Needs prompt?	Table	Equation	Handwriting	Two columns	Multiple columns
Marker		Markdown	N	✅ (markdown)	✅	✅	✅	❌
MonkeyOCR		Markdown	Y	✅ (html)	✅	✅	✅	✅
Nougat		Markdown	N	❌	✅	✅	✅	❌
MinerU		Markdown	N	✅ (html)	✅	❌	✅	❌
Llamaparse (balanced mode)	-	Markdown	Y	✅ (markdown)	❌	❌	✅	❌
Llamaparse (premium mode)	-	Markdown	Y	✅ (markdown)	❌	❌	✅	❌
Docling		Markdown	N	✅ (markdown)	❌	❌	✅	✅
RolmOCR		Markdown	Y	✅ (markdown)	✅	✅	✅	†
olmOCR		Markdown	Y	✅ (markdown)	✅	✅	✅	†
Unstructured		Raw text	N	❌	❌	❌	❌	✅
Pytesseract		Raw text	N	❌	❌	❌	✅	✅
MarkItDown		Markdown	N	❌	❌	❌	✅	✅
Amazon textract	-
Azure AI Document Intelligence	-
Google Cloud OCR	-
Mathpix	-
MistralOCR	-
Upstage	-
OmniAI	-
ChatDoc PDF parser	-
Reducto	-
OCRFlux
Nanonets
PaddleOCR
ClovaOCR	-
ParseExtract	-
Tensorlake	-
Vectorize	-
MassivePix	-
Dolphin
GOT
Manga OCR
EasyOCR
PDFeditify	-

† Process took too long

Layout Parsers

Usually outputs as JSON containing bounding box coordinates, content (as raw text or markdown), and sometimes type (header, figure, paragraph, etc.)

🚧 WORK IN PROGRESS

Models	Source	Output	Table	Equation	Handwriting	Two columns	Multiple columns
Chunkr
GroundX	-
ChatDOC	-
Unstract

Contributing

If you would like to contribute in any way, please read CONTRIBUTING.md and then make a contribution. Thank you!