Chunkr

Transforms complex documents into structured chunks for RAG and LLM applications. Parse PDFs, images, and spreadsheets into LLM-ready HTML/Markdown or JSON. OCR, layout detection, reading order, bounding boxes, citations, and schema-based extraction. It handles PDFs, DOCX, PPT, Excel, PNG, and JPEG files for versatile document processing. Tables receive structured extraction with preserved borders and cells, though shaded ones may output as images. A free tier allows experimentation, with open source access via GitHub for local deployment. It supports integration with models like Ollama through configurable YAML setups. Semantic chunking and bounding boxes ensure context aware splits optimized for retrieval accuracy. A 50 page PDF typically processes in under two minutes, scaling with document complexity.