Skip to main content

Getting Started

This page walks through the first 15 minutes — from a freshly imaged GPU server to a working Hub dashboard with at least one model loaded.

:::note Status This page is a stub. Full installation walkthrough coming with the first public release of the Hub installer. :::

Prerequisites​

  • A supported GPU server. NVIDIA DGX Spark (GB10) is the reference platform; any modern NVIDIA workstation or rack server with at least 64 GB system RAM and a recent driver will work.
  • A clean Linux install — Ubuntu 22.04 LTS or Ubuntu 24.04 LTS recommended.
  • A user account with sudo access.
  • A Tailscale account (free tier is fine). Optional but strongly recommended — it removes most of the public-DNS and TLS configuration steps.

1. Run the installer​

curl -fsSL https://install.serveraihub.ai | bash

The installer is idempotent and safe to re-run.

2. Open the dashboard​

After the installer finishes, open https://<your-hostname>.<tailnet>.ts.net in your browser. You'll see the Server AI Hub dashboard — the main view shows live GPU, memory, disk, and container status.

3. Sign in to integrations​

The first time you open the dashboard, you'll be prompted to connect:

  • GitHub — for pushing/pulling repositories from the Finder
  • Hugging Face — for downloading models without re-entering tokens
  • Optional: Slack / Discord — for alert notifications

Each connector is one click and survives reboots.

4. Load your first model​

From the Models tab in the top nav, click any model to see its size, license, and recommended hardware. Click Load and the Hub downloads + starts it. The container appears in the Containers tab with a status badge.

5. Open the chat​

The chat drawer (top-right) connects to LiteLLM, which routes to whichever models you have loaded. Try a query to confirm the loop works.

That's the install. From here, the User Guide walks through each feature one at a time.

Troubleshooting​

SymptomFix
Installer fails on docker composeRe-run the installer; it pins the Docker version and retries.
Dashboard 502 after installWait 60s — services boot in stages. If still down at 5 min, check the Hub container log via docker logs serveraihub-dashboard.
Tailscale hostname not resolvingOpen the Tailscale admin console; make sure the box has accepted the magic-DNS prompt.

If you hit something not covered here, email support@6ixlabs.ai.