# zylon-ai/private-gpt

Complete API layer for private AI applications on local models: RAG, skills, tools, MCP, text-to-sql, and more. Works with any OpenAI-compatible inference server.

Repository: https://github.com/zylon-ai/private-gpt
Canonical: https://ross.abutalabs.com/products/private-gpt
Homepage: https://www.zylon.ai/private-gpt
Language: Python
License: Apache-2.0
License Family: permissive
Topics: ai, ai-tools, on-premise
Last push: 2026-08-25T14:31:59+00:00

## Health v2 (maintenance only)
Score: 93/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 89, longevity 87
- inputs: {"age_days": 1219, "days_push": 8, "days_rel": 76, "gap_med": 14, "n_releases_24m": 2}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 57471, forks 7611 (observed 2026-08-28T04:12:18.537852+00:00)

## What it is
PrivateGPT is an open-source Python API layer that turns local or OpenAI-compatible LLM servers into production AI applications, providing RAG, tools, MCP connectors, text-to-SQL, and a messages API. It does not run models itself but connects to any inference server implementing /v1/chat/completions, and ships a built-in workbench UI for testing.

## Use cases
- build a private chatbot over my own documents
- run RAG on local models without cloud APIs
- expose an OpenAI-style API on top of Ollama
- add file ingestion and citations to an AI app
- connect local LLMs to custom tools and MCP servers
- query databases and CSVs with natural language
- deploy an on-premise AI assistant for sensitive data

## When to choose
- you need private, on-premise AI with no cloud dependency
- you already run an OpenAI-compatible server like Ollama, llama.cpp, or vLLM
- you want production-tested RAG, tools, and MCP building blocks behind one API
- you are building an AI product and want to avoid rebuilding backend primitives

## When to avoid
- you only need raw model inference without higher-level features
- you want a fully managed cloud AI service
- you need a polished end-user workspace rather than an API layer (consider Zylon)
- your stack is not Python-friendly

## Facets
- artifact type: framework
- maturity: active
- function: rag, agent-framework, llm-inference, api-framework, mcp, chatbot
- domain: artificial-intelligence, large-language-models, privacy, self-hosted, developer-tools
- platform: python, self-hosted, windows, cross-platform
- tags: local-llm, openai-compatible, on-premise, private-ai, ollama, text-to-sql, citations, api-layer, retrieval-augmented-generation, ai-agents, docker, linux, macos

## Member repositories
- zylon-ai/private-gpt (main) score 93

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:12:18.537852+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T16:18:30.409358+00:00, confidence not recorded.
  - readme: https://github.com/zylon-ai/private-gpt (fetched 2026-08-28T04:12:18.537852+00:00, sha b06940c403dc)
  - homepage: https://www.zylon.ai/private-gpt (fetched 2026-08-28T18:02:54.416037+00:00, sha 3e670550fe5e)
  - site_page: https://docs.zylon.ai/en (fetched 2026-08-28T18:02:54.426548+00:00, sha 6b6031b456f4)
  - site_page: https://www.zylon.ai/company/about (fetched 2026-08-28T18:02:54.428187+00:00, sha f2d53d13e094)
  - site_page: https://www.zylon.ai/platform/overview (fetched 2026-08-28T18:02:54.424694+00:00, sha 8b77dd0947d7)
- Data as of 2026-08-30T08:39:29.467469+00:00.