# Secure and Community

> Two tiers, one dedicated VM per session. Where each runs, who controls the hardware, and which one fits your work.

Every session on Node Lodge runs in a virtual machine of its own, with the GPU attached to it alone. The tier decides whose hardware sits underneath.

## At a glance

| | Secure | Community |
|---|---|---|
| Hardware | Partner datacenters | Machines of independent providers |
| Your session | A dedicated VM | A dedicated VM |
| GPU | Attached to your VM only | Attached to your VM only. Never shared. |
| Disk | Deleted at the end | Encrypted, erased at the end |
| Access | SSH, and JupyterLab through SSH | SSH, and JupyterLab in the browser |
| Inbound | SSH from your IP only | No port open on the provider's machine |
| Physical access | Datacenter controls | The provider keeps physical and administrator access |
| Session memory | Datacenter controls | Readable by the provider, GPU included |
| Best for | Confidential code and data | Open data, research, rendering, open models |

## Choose Secure when

- your code, data or model weights are confidential,
- a customer or a contract requires datacenter controls.

## Choose Community when

- your data is open or already public,
- you train, render or run open models and want the owner's price,
- you want JupyterLab in the browser without an SSH connection.

## What stays the same

- One account, one balance: credit covers both tiers. See [Billing](https://nodelodge.com/docs/billing).
- The same SSH key flow. See [Connect over SSH](https://nodelodge.com/docs/connect-ssh).
- Billed by the minute, at the rate shown when the session starts. See [Sessions and limits](https://nodelodge.com/docs/sessions-and-limits).

## For providers

Community machines belong to providers like you. Each one is reviewed before it is listed, and every session on it runs in its own VM: renters never reach your system. Start with [Provider requirements](https://nodelodge.com/docs/provider-requirements), then [Pair a machine](https://nodelodge.com/docs/pair-a-machine).