PrivateGPT — локальный RAG без передачи данных

★ 57.4K

PrivateGPT is an open-source API layer over local models that lets you build AI applications without data leaving your perimeter. Reach for it when security rules out the cloud but the application itself is ordinary: chat with documents, search over an internal knowledge base, processing incoming requests. It connects to any OpenAI-compatible inference server — Ollama, llama.cpp and others — and exposes the familiar interface, so your application code barely differs from code written against a cloud model. Ready-made pipeline pieces are included: document ingestion, chunking, vector retrieval, and answer assembly with source citations. It is not an end-user application but a foundation to build on: the interface and business logic remain yours to write.