RAG Anything
Basic Information
RAG-Anything is an all-in-one multimodal Retrieval-Augmented Generation (RAG) system and Python library for ingesting, parsing, indexing and querying documents that combine text, images, tables and mathematical expressions. It provides a unified pipeline from document ingestion through adaptive parsers (MinerU or Docling), modality-aware content analysis, multimodal knowledge graph construction, and hybrid retrieval. The project exposes programmatic APIs, configuration options and examples for end-to-end processing, direct insertion of pre-parsed content lists, batch processing, and integration with external LLM and vision model functions. It is intended for developers and researchers who need a single framework to convert heterogeneous documents into structured multimodal entities, preserve document hierarchy, and perform contextual queries that may include VLM-enhanced image analysis.
Links
Categorization
App Details
-
$$$$ -
$$$$