IBM open-source deep document parser converting complex PDFs, Office files into Markdown and JSON.
Licensed under MIT for 100% free personal and commercial use with zero local processing caps.
Docling by IBM is a state-of-the-art document conversion engine engineered for LLM and RAG pipelines. It parses PDF, DOCX, and PPTX documents with granular layout understanding, accurately extracting multi-column passages, mathematical LaTeX formulas, complex nested tables, and reading orders into clean Markdown and structured JSON.