Yuliang-Liu/MonkeyOCR

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

6635 stars · 461 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

MonkeyOCR is a lightweight large multimodal model (LMM) for document parsing that uses a Structure-Recognition-Relation triplet paradigm to convert documents (including PDFs and images) into structured text like Markdown. It provides pretrained model weights, a Python inference pipeline, and a demo, with an active v2 release line.

Use cases

• parse pdf documents into markdown
• extract text and tables from scanned documents
• ocr for academic papers with formulas
• convert document images to structured text
• multilingual document parsing
• batch document digitization pipeline

When to choose

• you need open-source document parsing with layout, text, formula, and table recognition
• you want a lightweight multimodal OCR model with pretrained weights
• you need multilingual document extraction in Python

When to avoid

• you need only plain OCR without layout or structure understanding
• you have no GPU and need fast lightweight processing
• you need a fully managed cloud OCR API

Facets

library · maturity active

ocr pdf machine-learning llm-inference image-processing artificial-intelligence computer-vision pdf deep-learning python cross-platform document-parsing multimodal vision-language-model layout-analysis pdf-to-markdown natural-language-processing gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="Yuliang-Liu/MonkeyOCR")