olmOCR
Open-source toolkit that uses a vision language model to turn PDFs and scans into Markdown.
These buttons open the developer's own site, repository or store listing in a new tab. wares.gg does not host downloads.
About olmOCR
olmOCR converts PDFs, PNG and JPEG documents into clean Markdown with a natural reading order, even with multi-column layouts, figures and insets. It handles equations, tables and handwriting and removes headers and footers.
It runs a 7B parameter vision language model, so it needs a GPU, and Docker images are provided. It is aimed at people processing large document collections, such as for LLM datasets. An online demo is available for trying it without setup.
Strengths
- Handles tables, equations and handwriting
- Keeps reading order in complex layouts
- Docker images available
- Model and training code are open
Limitations
- Requires a capable GPU
- Command-line pipeline with no desktop interface
Details
- Pricing
- FreeFree and open source under the Apache 2.0 licence; you supply the GPU.
- License
- Apache-2.0
- Developer
- Allen Institute for AI
- Platforms
- Linux, Command line
- How it runs
- Downloadable app
- Account
- Not required
- Works offline
- Yes
- Best suited for
- Converting large batches of PDFs into clean text on your own GPU
- Categories
- PDF tools, Local AI tools
- Last verified
- Added
- Provenance
- Facts checked against the developer's own pages and store listings, 2 sources on file.
Alternatives to olmOCR
Compare allSoftware that can replace olmOCR for an important use case, and what changes if you switch.
Marker
An open-source tool that converts PDFs and other documents to Markdown, JSON and HTML.
Marker also converts PDFs to Markdown with tables and equations, runs on CPU, GPU or Apple MPS, but its model weights have a separate licence.
MinerU
A document parser that converts PDFs, Office files and scans into Markdown and JSON.
MinerU converts PDFs, Office files and scans to Markdown and JSON with several local backends from ONNX to vLLM, under a non-standard licence.
PaddleOCR
OCR toolkit that turns PDFs and images into structured data.
PaddleOCR parses layout, tables and formulas into JSON and Markdown on Windows, macOS and Linux, built on PaddlePaddle with mostly Chinese documentation.
MarkItDown
Python command-line tool that converts PDFs, Office files and other documents into Markdown.
MarkItDown is a lightweight MIT-licensed Python tool converting PDFs and Office files to Markdown without a GPU, but it is not aimed at high-fidelity or scanned input.
Tesseract OCR
The open-source OCR engine behind most free text recognition tools.
Tesseract is a classic OCR engine with over a hundred languages that outputs text or searchable PDFs, but not Markdown, and needs manual image preprocessing.
EasyOCR
Python OCR library covering more than eighty languages out of the box.
EasyOCR is a Python OCR library covering eighty languages that returns text with bounding boxes, not structured Markdown, and is slow without a GPU.
Similar software
Related functionality, not necessarily a direct replacement.
PDFMathTranslate
An AI tool that translates scientific PDFs while keeping formulas, charts and layout intact.
Poppler
PDF rendering library with pdftotext, pdftoppm and pdfinfo tools.
K2pdfopt
A tool that reflows and resizes PDF and DjVu files for reading on e-readers and phones.
