baidu /
Unlimited-OCR
Open OCR model that turns images and multi-page PDFs into structured text in one pass
- 28.2k stars
- 1.6k stars since 3 Oct
- Python
- Updated 10 days ago
- Some experience
- Models & local AI
Classic OCR tools read one page at a time and often lose layout, tables and reading order. Unlimited-OCR is a model that looks at page images and writes out the parsed document, and it is built to handle long inputs such as many pages in a single request.
You give it an image or a list of page images with a short prompt like 'document parsing.' or 'Multi page parsing.' For a single image there are two modes, 'gundam' (cropped, 640 px) and 'base' (1024 px). Multi-page and PDF input uses base mode only. A PDF is first turned into page images with PyMuPDF, then parsed together.
The README says it aims to push DeepSeek-OCR one step further and credits DeepSeek-OCR, DeepSeek-OCR-2 and PaddleOCR. It ships with a no-repeat n-gram setting to stop the model looping on long outputs. The code is MIT licensed and the weights are on Hugging Face and ModelScope.
What it does
Multi-page and PDF parsing
Pass several page images, or a PDF converted to images, and get one parsed result for the whole document.
Two single-image modes
Choose 'gundam' (cropped, image size 640) or 'base' (image size 1024) for a single image.
Three ways to run it
Use Hugging Face transformers on an NVIDIA GPU, a vLLM Docker image, or an SGLang server with an OpenAI-compatible API.
Batch script
infer.py starts an SGLang server and sends concurrent requests for a folder of images or a PDF.
Training and hosting options
The release notes say it supports training with ms-swift and is available on Baidu Cloud, ModelScope and a Hugging Face demo.
What you could build
- Convert a scanned PDF report into parsed text in a single multi-page run.
- Batch-process a folder of document images with infer.py and a few concurrent requests.
- Serve an OCR endpoint with vLLM or SGLang and call it from your own app.
- Compare its output against DeepSeek-OCR on your own documents.
- Fine-tune the model on your own document types with ms-swift.
Get started
Set up Python 3.12 with CUDA and the pinned packages listed in the README (torch, transformers, PyMuPDF and others). Then load baidu/Unlimited-OCR with transformers and call model.infer on an image or model.infer_multi on page images. For serving, follow the vLLM recipe or the SGLang steps.
docker pull vllm/vllm-openai:unlimited-ocrdocker pull vllm/vllm-openai:unlimited-ocr-cu129Commands are copied from the project's README. Read it on GitHub for the full setup.
More in Models & local AI
transformers
Shared model definitions for text, vision, audio and multimodal AI models
open-webui
Self-hosted chat interface for Ollama and OpenAI-compatible models, with RAG and plugins
odysseus
Self-hosted workspace for AI chat, agents, research, documents, email and calendar
minimind
Train a tiny 64M-parameter language model from scratch on one consumer GPU
Kronos
Pre-trained Transformer models that forecast candlestick (K-line) price data
airllm
Run very large open language models on a small GPU by loading one layer at a time
Vane
Self-hosted AI answer engine that searches the web and cites its sources
timesfm
Pretrained Google model that forecasts time series without training on your data
gallery
Android and iOS app for running open LLMs offline on your phone
lingbot-map
Feed-forward model that builds 3D scenes from video frames as they stream in