What is PrivateGPT used for?

PrivateGPT is an open-source API layer that adds RAG, tools, MCP connectors and a built-in test UI to locally hosted models for private AI apps. It is a good fit for private RAG applications on local models, adding tools and MCP to local LLMs, and on-premise AI products without cloud APIs.

Is PrivateGPT open source?

Yes. PrivateGPT is open source under the Apache-2.0 licence. Its source code is on GitHub at zylon-ai/private-gpt and is written mainly in Python.

Yes. PrivateGPT is open source, so the software itself is free to use.

Can I self-host PrivateGPT?

Yes. PrivateGPT can be self-hosted on your own server or infrastructure.

What is PrivateGPT an alternative to?

PrivateGPT is an open-source alternative to ChatGPT. Other open-source alternatives to ChatGPT include Ollama, LocalAI and InternVL.

Is PrivateGPT actively maintained?

Yes. The most recent commit to PrivateGPT was on 2 October 2026, and the latest release is v1.0.1, published on 18 June 2026. The project has 58k stars on GitHub.

PrivateGPT: open-source alternative to ChatGPT

About PrivateGPT

PrivateGPT is an open-source API layer that sits between your applications and a locally hosted language model. Running a model is only the first step, so the project supplies the higher-level building blocks, such as retrieval, tools and file ingestion, that applications usually need. It is written in Python, released under the Apache-2.0 license, and used to power the Zylon on-premise AI platform.

The project leaves model execution to other software: it links to whichever inference server speaks the OpenAI API, such as Ollama, through a configurable base URL. On top of that it offers a messages API (streaming, async and token counting), ingestion of files and artifacts, retrieval that cites its sources, agentic RAG, built-in tools for web search, page fetching and running code, plus user-defined tools, MCP connectors, database and CSV access, and embeddings.

A built-in workbench UI is available for testing and demos, while the API is described as the real product. Quickstart instructions cover Docker and other install options. It suits developers and enterprises that want to build AI products without sending data to cloud APIs, and that already run or plan to run a local inference server.

Key features

Messages API with streaming and token counting
Retrieval with citations and agentic RAG
File and artifact ingestion
Built-in web search, web fetch and code execution tools
Custom tools and MCP connectors
Works with any OpenAI-compatible inference server
Built-in workbench UI for testing

Good fit for

→Private RAG applications on local models
→Adding tools and MCP to local LLMs
→On-premise AI products without cloud APIs

Built with: Python
Tags: private-ai; rag; local-llm; api; mcp; on-premise; python; self-hosted

PrivateGPT: questions and answers

What is PrivateGPT used for?: PrivateGPT is an open-source API layer that adds RAG, tools, MCP connectors and a built-in test UI to locally hosted models for private AI apps. It is a good fit for private RAG applications on local models, adding tools and MCP to local LLMs, and on-premise AI products without cloud APIs.
Is PrivateGPT open source?: Yes. PrivateGPT is open source under the Apache-2.0 licence. Its source code is on GitHub at zylon-ai/private-gpt and is written mainly in Python.
Is PrivateGPT free?: Yes. PrivateGPT is open source, so the software itself is free to use.
Can I self-host PrivateGPT?: Yes. PrivateGPT can be self-hosted on your own server or infrastructure.
What is PrivateGPT an alternative to?: PrivateGPT is an open-source alternative to ChatGPT. Other open-source alternatives to ChatGPT include Ollama, LocalAI and InternVL.
Is PrivateGPT actively maintained?: Yes. The most recent commit to PrivateGPT was on 2 October 2026, and the latest release is v1.0.1, published on 18 June 2026. The project has 58k stars on GitHub.