OmniParse is an open-source data ingestion and parsing platform that converts documents, images, audio, video, and web pages into clean, structured output optimized for generative AI applications. It has earned more than 7,900 stars on GitHub and is licensed under GPL-3.0, with development activity continuing through late 2025.
OmniParse is a platform that ingests and parses any unstructured data into structured, actionable data optimized for GenAI (LLM) applications. Whether you are working with documents, tables, images, videos, audio files, or web pages, OmniParse prepares your data to be clean and LLM-ready. It bundles specialized models behind one ingestion API: OCR and vision transformers for documents and images, a Whisper-based API for audio transcription, and a web crawler for pages. Outputs are normalized into structured formats that slot directly into retrieval-augmented generation pipelines and other LLM workflows, sparing teams the usual patchwork of one-off parsers. The project is self-hostable, so data stays under your control, and its parser-library design lets developers call individual components without adopting the whole server.
OmniParse is free and open source under the GPL-3.0 license. You can self-host the full platform at no cost beyond your infrastructure. There is no per-document pricing; heavy workloads simply scale with the compute you provide.
OmniParse is for developers building GenAI applications on messy real-world data: RAG systems over mixed document collections, enterprise search, and data teams that need one consistent parsing layer instead of maintaining separate OCR, transcription, and scraping code paths.