# Chunkr

> Document intelligence API that converts complex documents into structured, LLM-ready data with OCR, layout analysis, and chunking.

Chunkr, made by Lumina AI Inc. (dba Chunkr), is a document intelligence API that turns PDFs, Office files, spreadsheets and images into structured, RAG/LLM-ready output. It offers a managed Cloud API and an AGPL-3.0 open-source version that can be self-hosted. The homepage notes that Chunkr is joining Extend, with a migration guide published in its docs.

## What It Is

Chunkr is infrastructure for document parsing and extraction. Its Parse workflow returns HTML or Markdown plus bounding boxes at segment, line and word level, and its Extract workflow turns documents into structured data that follows a JSON Schema, with citations and confidence scores. Output includes RAG-optimized chunks, image crops and file metadata.

## Open Source vs Cloud

The README says the open-source release uses community models, while the Cloud API runs proprietary in-house models (layout-1, parse-1, order-1) and adds native Excel support. The vendor describes these models as more accurate; the accuracy figures on its site are vendor-published claims. Enterprise deployment can be on-prem or VPC, according to the vendor.

## Setup Path

The open-source version runs through Docker Compose, with GPU, CPU-only and Mac ARM variants, and serves a web UI and an API locally. LLMs are configured via a models.yaml file or environment variables, using any OpenAI-compatible endpoint, such as OpenAI, Google AI Studio, OpenRouter, vLLM or Ollama. The Cloud API offers TypeScript and Python SDKs and a playground.

## Security and Compliance

The vendor states it is SOC 2 and HIPAA compliant, with configurable zero data retention.

## Features
- Layout analysis
- OCR with word-level bounding boxes
- HTML and Markdown output
- Structured extraction with JSON Schema compliance
- Citations and confidence scores
- Reading order detection
- RAG-optimized chunking
- Support for PDF, Word, PowerPoint, images and spreadsheets
- Self-hosting via Docker Compose
- Configurable LLM providers via models.yaml

## Integrations
OpenAI, Google AI Studio, OpenRouter, vLLM, Ollama, Docker Compose

## Platforms
API, WEB, DEVELOPER_SDK, LINUX, MACOS

## Pricing
Open Source, Free tier available

## Version
v2.2.1

## Links
- Website: https://chunkr.ai
- Documentation: https://docs.chunkr.ai/
- Repository: https://github.com/lumina-ai-inc/chunkr
- EveryDev.ai: https://www.everydev.ai/tools/chunkr