OpenRAG vs PrivateGPT

OpenRAG is a modular framework for building systems that answer questions using documents as their grounding. Its retrieval combines semantic search with BM25 keyword matching and multilingual reranking; answers cite a source document and page, with links to the cited location. Document processing covers listed text, office, email, audio, video, and image formats, with PDF layout awareness, OCR, image captioning, and audio transcription. The product includes an admin console, chat interface, and OpenAI-compatible API, and can connect to user-provided models or a hosted provider. Partition isolation, owner, editor, and viewer roles, plus token or OpenID Connect authentication, are among its access controls. It can run on infrastructure controlled by the deployer, keeping documents, embeddings, and queries within that environment. The self-hosted AGPL-3.0 edition is listed at free. Deployment prerequisites are Docker, Docker Compose, and at least 16 GB of RAM; a CPU-only profile is optional. CSV, ODT, and HTML support, tool calling, agentic RAG, MCP, and encryption in transit and at rest are listed as coming soon.

PrivateGPT is a free, open-source API layer for building private AI applications on your own infrastructure. It does not run models itself: it connects to a separate OpenAI-compatible inference server, which can be local or remote. The Messages API supports streaming, asynchronous processing, and token counting. Document workflows include file and artifact ingestion, retrieval with citations, and tools for querying databases and analyzing tabular data. The documented capabilities also include web search and extraction, code execution, custom tools, and MCP connectors. A browser workbench at `/ui` can be used for demos and inspecting API requests, but it stores chats and settings in browser localStorage and has no access control; the documentation describes it as a local demonstrator. PrivateGPT can be installed as a Python package, from source, or with Docker, and runs as a local server. Package installation requires Python 3.11; Python 3.10 and 3.12 or later are unsupported. Platforms include Linux, macOS, Windows, web, API, and self-hosted environments.