Should you self-host?
AxonOS is open source (MIT), so you can run it on your own hardware. For most researchers the hosted platform at app.axonos.io is the easiest path — but self-hosting makes sense in some cases.
If you just want to use AxonOS, start with Launch AxonOS instead. The pages in this track are for people running their own instance.
Hosted vs. self-hosted
| Hosted (app.axonos.io) | Self-hosted | |
|---|---|---|
| Setup | None — open a browser | Install Docker, build the image |
| GPUs | Provided, billed per minute | Your own hardware |
| Payments & billing | Built in (wallet + credits) | Optional — only with the gated stack |
| Maintenance | Managed for you | You operate it |
| Best for | Researchers, students, quick work | Labs, classrooms, operators, contributors |
Choose your path
🖥️A single desktop
One docker run gives you a personal AxonOS desktop on port 6080 — great for local,
offline, or classroom use. No wallet or billing required.
🌐The full gated platform
Run the same wallet-gated, multi-session stack that powers app.axonos.io — gate, launcher, Postgres, and TURN — with Docker Compose.
Docker Compose →🧰A custom build
Use the CLI/GUI launcher to pick exactly which scientific apps and AI models to bake in.
The Launcher →🔌Extend it
Add your own tools with the YAML/JSON plugin system and template builder.
Plugins →Before you start
Check the requirements
Confirm Docker, and (optionally) an NVIDIA GPU + Container Toolkit. See System Requirements.
Pick single-desktop or full stack
A plain docker run for one desktop, or Docker Compose for
the gated platform.
Configure your environment
Copy env.example to .env and set your secrets. See
Configuration.
Harden before exposing it
If it's reachable beyond localhost, read Security & Self-Hosting first.
How it works under the hood
Curious what you're running? Architecture maps the components, and User Flow & Gateway walks the request lifecycle. The economic model lives in Payment Rails & Tokenomics.
Everything is on GitHub at github.com/AXDT-INC/AxonOS.