#1616th of 24 in Data and scraping for AIFor data and documents, documents and ocr
olmocr
PDF and image to Markdown converter built on a 7B vision-language model
Live demo ↗
(opens in a new tab)Website ↗
(opens in a new tab)Repository on GitHub ↗
(opens in a new tab)
- Stars
- 19.8k
- License
- Apache-2.0
- Last commit
- Mar 2026
- Last release
- Mar 2026
- Language
- Python
Missed the latest GitHub check; these facts may be older.
Overview
olmOCR converts PDFs, PNGs and JPEGs into Markdown using a 7B vision-language model served through vLLM. It handles equations, tables, handwriting and multi-column layouts, and strips headers and footers. It runs on a local NVIDIA GPU, against a remote OpenAI-compatible server, or across many nodes coordinated through an S3 workspace.
Who it is for: Engineers batch-converting PDFs and scans to Markdown for LLM pipelines
Strengths
- Scores 82.4 on olmOCR-Bench (v0.4.0), near the top of the README table
- Accepts any OpenAI-compatible server via --server, so no local GPU is required
- Multi-node batch mode with an S3 work queue for millions of PDFs
- Ships benchmark suite and trainer code for reproducing or fine-tuning the model
Weaknesses
- Local inference needs an NVIDIA GPU with at least 12 GB VRAM and 30 GB disk
- README advises a clean conda environment because dependencies are hard to install
- System setup needs poppler-utils and extra fonts; instructions cover only Ubuntu/Debian
- Docker image with the model is about 30 GB
What it needs
- GPU required
- Docker
- Needs vLLM, poppler-utils, NVIDIA GPU (12 GB+ VRAM), AWS S3 (optional, multi-node)
- Models: olmOCR-2-7B-1025-FP8, olmOCR-7B-0825-FP8, olmOCR-7B-0725-FP8
Also in Data and scraping for AI
See all 24| Rank | Project | Score |
|---|---|---|
| 1 | mineruDocument parser that converts PDFs, images and Office files to Markdown and JSON | 81 out of 100 |
| 2 | Crawl4AIPython web crawler that converts pages to LLM-ready Markdown and JSON | 79 out of 100 |
| 3 | doclingPython library and CLI that converts PDFs, Office files and more into structured text | 79 out of 100 |
| #4 | label-studioMulti-user data labeling tool for audio, text, images, video and time series | 74 out of 100 |
| #5 | markitdownPython tool that converts PDF, Office, HTML and audio files to Markdown for LLMs | 71 out of 100 |