OCRmyPDF vs Granite Docling
Pick OCRmyPDF to make scanned PDF archives searchable and keep them as PDF/A. Pick Granite Docling to convert English PDFs into structured output such as Markdown, HTML or LaTeX, including tables, code and equations. The main split is the output: OCRmyPDF adds a text layer to the same PDF, while Granite Docling rebuilds the page structure.
OCRmyPDFocrmypdf/ocrmypdf | Granite Doclingibm/granite-docling | |
|---|---|---|
| Type | Open source | Open source |
| Price | 1 credit / page | 1 credit / page |
| Per 1,000 pages | $1.00 | $1.00 |
| Typical speed | 1.0 s / page | 1.5 s / page |
| Best for | Searchable PDFs from scans | Small-model document conversion |
| Inputs | PDF, images | PDF, images |
| Languages | 100+ languages (via Tesseract) | English, experimental CJK and Arabic |
| Max pages per request | 1,000 | 500 |
| Tables | No | ✓ Yes |
| Handwriting | No | No |
| Math | No | ✓ Yes |
| Bounding boxes | No | ✓ Yes |
| Confidence | No | No |
| Forms | No | No |
| Multilingual | ✓ Yes | No |
| On OCRRouter | Available | Available |
Prices are what you pay on OCRRouter. Speed is the provider's typical figure, not a benchmark we ran.
Which to choose
Choose OCRmyPDF if
- You need validated searchable PDF/A output for archiving.
- You want the OCR layer inserted losslessly where possible, without re-encoding other content.
- Your documents are not English only, since Tesseract language packs cover 100+ languages.
Choose Granite Docling if
- You need Markdown, HTML, LaTeX, DocTags or bounding boxes instead of a searchable PDF.
- Your English PDFs contain tables, code or equations: the model card reports table structure TEDS of 0.97 on FinTabNet, code F1 of 0.988 and equation F1 of 0.968.
- You already run Docling and want a small 258M parameter model that runs on CPU or a small GPU under Apache-2.0.
Key differences
| OCRmyPDF | Granite Docling | |
|---|---|---|
| What it does | Adds a Tesseract OCR text layer to scanned PDFs | Document conversion VLM that outputs DocTags for the Docling library |
| Output formats | Searchable PDF (PDF/A by default), plain text | DocTags, Markdown, HTML, LaTeX, bounding boxes |
| Licence and price | MPL-2.0, free, self-hosted (you pay for compute) | Apache-2.0, free, self-hosted (you pay for compute) |
| Hardware | CPU only, spreads work across all CPU cores | 258M parameters, runs on CPU or a small GPU |
| Languages | 100+ via Tesseract language packs | English primary; Japanese, Arabic and Chinese are experimental |
| Tables and layout | No layout, table or Markdown output | Table structure TEDS 0.97 on FinTabNet (150dpi, authors' figure) |
| Known failure | File size can double or quadruple with --deskew --clean-final (issue #948); --skip-text with default PDF/A output removes form fields (issue #1790); hangs on PDFs whose forms share one Resources dictionary (issue #1763) | Repetitive loops on page headings and invoices; may output only exclamation marks on GPUs without bfloat16 such as T4 unless run in float32 |
Sources
- OCRmyPDF GitHub README
- Issue #948 file size
- Issue #1790 form fields
- Granite-Docling model card
- HF discussion #24 repetition
What each one is
OCRmyPDF
Command-line tool that adds a searchable text layer to scanned PDFs, using Tesseract.
OCRmyPDF detailsGranite Docling
Compact IBM vision model that converts pages to DocTags for the Docling pipeline.
Granite Docling detailsTry both on your own file
Run the file once with each model and compare the output.
PDF, PNG, JPG, WebP, HEIC. Up to 20 MB each, 10 files.No file?
Use one, fall back to the other
{"model": "ocrmypdf/ocrmypdf","fallbacks": ["ibm/granite-docling"],"file_url": "https://example.com/document.pdf"}Questions
More comparisons
OCRmyPDF vs SmolDoclingOCRmyPDF vs Qwen-VLRapidOCR vs Granite DoclingEasyOCR vs Granite DoclingGoogle Document AI vs Granite DoclingAzure Document Intelligence vs Granite DoclingAzure Document Intelligence vs OCRmyPDFJina OCR vs OCRmyPDF
OCRmyPDF alternativesGranite Docling alternativesBest OCR models in 2026