#1717th of 20 in Data and scraping for AIFor data and documents
chandra
OCR model that converts PDFs and images to Markdown, HTML or JSON with layout
Live demo ↗
(opens in a new tab)Documentation ↗
(opens in a new tab)Website ↗
(opens in a new tab)Repository on GitHub ↗
(opens in a new tab)
- Stars
- 12.5k
- License
- Apache-2.0
- Last commit
- Jun 2026
- Last release
- Mar 2026
- Language
- Python
7,958 on 31 Mar 202612,521 on 20 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Datalab is hiring: 5 open roles
Overview
Chandra OCR 2 is a vision-language OCR model and Python CLI that turns images and PDFs into Markdown, HTML or JSON while keeping layout, tables, math, forms and handwriting. It runs locally through HuggingFace or against a vLLM server, and ships a Streamlit demo app. Weights use a modified OpenRAIL-M license, separate from the Apache-2.0 code.
Who it is for: Engineers extracting structured text from PDFs, scans and handwritten documents
Strengths
- Outputs Markdown, HTML and JSON with layout, extracted images and captions
- Supports 90+ languages; README reports 77.8% average on its own 43-language benchmark
- Two backends: local HuggingFace or vLLM server with parallel workers
- Handles handwriting, math, tables and checkbox forms; CLI takes page ranges and directories
Weaknesses
- Weights are OpenRAIL-M: free only for research, personal use and startups under $2M
- Hosted Datalab API is stated to be more accurate than the open weights
- Benchmarks are mostly self-published; throughput is 1.44 pages/s on an H100
- The chandra_vllm launcher depends on Docker; VRAM and RAM requirements are unknown
What it needs
- GPU required
- Needs vLLM, Docker, HuggingFace Transformers, PyTorch
- Models: datalab-to/chandra-ocr-2, HuggingFace, vLLM
- port 8000
Also in Data and scraping for AI
See all 20| Rank | Project | Score |
|---|---|---|
| 1 | mineruDocument parser that converts PDFs, images and Office files to Markdown and JSON | 80 out of 100 |
| 2 | Crawl4AIPython web crawler that converts pages to LLM-ready Markdown and JSON | 78 out of 100 |
| 3 | doclingPython library and CLI that converts PDFs, Office files and more into structured text | 78 out of 100 |
| #4 | label-studioMulti-user data labeling tool for audio, text, images, video and time series | 72 out of 100 |
| #5 | markitdownPython tool that converts PDF, Office, HTML and audio files to Markdown for LLMs | 71 out of 100 |