Chunkr
An open-source document intelligence API that performs layout analysis, OCR and semantic chunking to convert PDFs, presentations, Word documents and images into structured data for RAG and LLM pipelines.
Chunkr is a service for document layout analysis, OCR and semantic chunking. It converts PDFs, PowerPoint files, Word documents and images into chunks suitable for retrieval-augmented generation and other LLM pipelines, producing structured HTML and Markdown output alongside bounding boxes from its OCR step.
The open-source release, licensed under AGPL-3.0, uses community and open-source models and runs through Docker Compose, with optional GPU support through the NVIDIA Container Toolkit. A separate hosted Cloud API and an Enterprise edition run proprietary in-house models and support on-premises or VPC deployment for higher accuracy and regulated environments.
It is written in Rust.