Install & Activate
Get AgentOS running in a few minutes. The packaged build includes everything it needs, so you don't have to install Node.js or npm.
Requirements
- Operating system: Windows 10/11 (64-bit) or Linux (x64).
- A model: a running Ollama or llama.cpp server, or an API key for an OpenAI-compatible provider. See Connecting a Model.
- An activation token: provided by your organization or generated from your Symbient AI account.
- A modern browser for the web interface.
The agent itself runs comfortably on commodity hardware. Hardware requirements mostly depend on the model you choose to run locally.
1. Download and extract
Download the archive for your platform and extract it to a permanent location, such as C:\AgentOS or ~/agentos.
| Platform | Package | Launcher |
|---|---|---|
| Windows 64-bit | agentos-win64-v<version>.zip | agentos.exe |
| Linux x64 | agentos-linux-v<version>.tar.gz | agentos |
To use agentos from any terminal, add the install folder to your PATH.
2. Start the server and open the web UI
Running agentos with no arguments starts the server if it isn't already running and opens the web UI in your browser:
agentos
Or control the server directly:
agentos server start # start the background server
agentos open # open http://localhost:3456
agentos server status # is it running?
agentos server restart # restart (applies config changes)
agentos server stop # stop the server
agentos server logs # tail the live server log
The server listens on port 3456 by default. You can change it in Global Settings.
3. Activate AgentOS
The first time you open the web UI, AgentOS asks for an activation token. It looks like AGOS-XXXXX-XXXXX-XXXXX-XXXXX.
- Get a token from your organization's administrator, or sign in to Symbient AI and generate one from your organization's license page.
- Paste it into the Activation Token field and click Activate AgentOS.
AgentOS stores a signed license certificate locally and re-verifies it when the server starts. If it can't reach the license server, it keeps working for up to 30 days. See Updates & Licensing.
4. Start automatically (optional)
Turn on Auto-start in Global Settings to launch the server when you log in, so scheduled tasks and remote access
keep working after a reboot. On Windows this registers a Task Scheduler task. On Linux it installs a systemd user
unit, and on macOS a launchd agent.
Next steps
Once activated, the setup wizard walks you through connecting a model and creating your first workspace: