pAIring overview

pAIring (Private AI ring) is a self-hosted platform for running private LLMs and image models. It works for one person, a team, a department or a whole organization. It has role-based access, load balancing with failover, and checks on risky settings so configuration mistakes are caught early.

It supports public GGUF models out of the box. You can browse them from Hugging Face or from your own private repository. Nothing leaves your network.

Under active development

pAIring changes often. New features and fixes arrive regularly, and some things described here may change. This documentation is updated along with the project.

Where to start

• Deployment layouts: one node, or separate inference servers. Read this first.
• Quick start: install and first steps.
• Features: everything pAIring can do.
• Chat and channels: every chat option, including notes, attachments and tools.
• Your settings: the settings every user has.
• Image generation: set up the image engine and create images.
• Admin settings: every admin tab explained.
• Configuration: environment variables and storage folders.
• Stats and monitoring: usage, telemetry and hardware.
• Connectors: run chats on a cloud GPU service.
• Security: how data is protected, and a checklist before going live.
• Matricxon docs: the inference engine that works best with pAIring.

Runs on

• Linux: all features are available.
• macOS: works. Install Ollama or Matricxon yourself and point pAIring at it.
• Windows: not supported at this point.
• Hardware: x86_64 or arm64. A GPU is optional. CPU-only works, and a GPU (NVIDIA or Apple Silicon) makes replies much faster.
• Memory: there is no fixed requirement. It depends on the models you run, and the Models settings page shows what fits your hardware.

Built to be simple

• Plain HTML, CSS and JavaScript with Tailwind’s browser build. There is no Node and no build step.
• SQLite out of the box, with MySQL when you need more.
• The web UI is a client of pAIring’s own REST API, with more than 90 endpoints. Live docs are at /docs (Swagger UI) and /redoc on your server.
• GET /health reports whether this instance can serve a chat request right now.

License

pAIring is released under GNU AGPL-3.0. You can run, modify and self-host it for personal or internal use. If you offer a modified version to others as a hosted service, you share those changes back.