Deploy on a Managed Container Platform

Use this path when the provider handles the Linux host, NVIDIA driver, and Docker runtime and asks you only for a container image and settings.

1. Create a GPU container

Choose a GPU and enough storage for the models you plan to use.

Choosing a GPU: More VRAM lets you run larger models — 16 GB handles most image generation, 24-48 GB covers mid-size LLMs, and 100 GB+ is needed for large LLMs.

Set the custom image to:

openlaboratoryorg/laboratory-os

2. Configure access and persistence

Add one remote-access credential or a supported combination:

Key	Value
`UPLINK_DEVICE_TOKEN`	Managed mode. A host-bound, serve-only token created in the Uplink console for the account that should own this Laboratory instance.
`UPLINK_EDGE`	Self-hosted mode. The Uplink edge domain and its static admission key. Use these together instead of a managed device token.
`UPLINK_EDGE_API_KEY`	The static admission key for the custom edge.
`TS_AUTHKEY`	Tailscale mode. Join the station to a tailnet. The auth key works on its own or alongside either Uplink mode; the hostname is optional.
`TS_HOSTNAME`	Optional hostname for the station on its tailnet.

Mount persistent storage at /workspace. A volume mounted elsewhere will not preserve installed apps, models, and Laboratory settings. Confirm whether the provider keeps that volume when the container is stopped, rescheduled, or deleted.

The container should run continuously rather than as a finite batch job, and the platform must expose its logs.

No inbound ports or firewall rules are needed — Laboratory OS uses an outbound tunnel. You can block all inbound traffic entirely.

3. Launch and connect

Start the container and wait for the image pull and station boot to complete.

Check the container logs for the assigned Laboratory URL. It appears once the selected remote-access method is ready, usually within the first 30 seconds:

docker logs laboratory

In device-token mode, the instance also appears under Tunnels and Devices in the Uplink console, and access is gated by the owning Uplink account. On a self-hosted edge, use the lab access token printed during boot. On Tailscale, access is governed by your tailnet ACLs.

4. Before stopping or moving it

Verify that /workspace is on durable storage.
Keep a backup if the provider can reclaim or detach the underlying machine.
Check whether stopped containers continue to incur GPU, disk, or reservation charges.
Reuse the same /workspace volume when recreating the container with a newer Laboratory OS image.

See Persistence for the storage contract and Run Laboratory OS for the complete environment-variable and image-tag reference.