---
title: "Register AI Compute with Brev — Instructions"
canonical: "https://build.nvidia.com/spark/brev/instructions.md"
---

# Step 1. Log in to Brev

Go to the [Brev UI](https://brev.nvidia.com), log in, and confirm you are in the correct org. Once logged in, open [Compute](https://brev.nvidia.com/org/environments).

Click **Connect a local device** (or **Register Compute**) and follow the instructions in the pop-up window.

# Step 2. Complete the pop-up instructions

In the Brev Connect flow:

- Add a name for the compute
- Create or enter an unexpired **Read & Write Personal API key** for your org
- Run the generated command on your hardware platform to install the Brev CLI and register the compute (requires administrative privileges)

Save your API key and click **Done** after registration. Do not share the key or command.

# Step 3. Enable SSH access

On your hardware platform, enable SSH and grant yourself access using the same API key:

```bash
export PATH="$HOME/.local/bin:$PATH"
brev enable-ssh --api-key "<your-api-key>"
brev grant-ssh --api-key "<your-api-key>"
```

Grant your Brev user access using the same Linux user and SSH port (default: `22`) selected during SSH setup.

# Step 4. Confirm registration in the Brev UI

1. Go to the [Brev UI](https://brev.nvidia.com)
2. Open [Compute](https://brev.nvidia.com/org/environments)
3. Confirm that your hardware platform appears as a registered node with a **Connected** status

# Step 5. Next steps

Your hardware platform is now integrated into Brev as a secure, remotely accessible GPU environment.

You can share access through the Brev UI by:

1. Adding the user to your [Team](https://brev.nvidia.com/org/team)
2. Opening your instance under [Compute](https://brev.nvidia.com/org/environments)
3. In the **SSH Access** section for the instance, search for the user, click **Modify Access**, enable access for the Linux user selected in Step 3, and click **Save**

On your local machine, [install the Brev CLI](https://docs.nvidia.com/brev/cli/getting-started), run `brev login`, select the same org with `brev org set <org>`, and connect with `brev shell <name>`.

# Step 6. Cleanup

To unregister your hardware platform and complete local cleanup, run the following on that platform.

**CLI:**

```bash
brev deregister --api-key "<your-api-key>"
```

**UI:**

UI removal revokes access. Run `brev deregister` on the hardware platform to complete local cleanup.

1. Go to the [Brev UI](https://brev.nvidia.com)
2. Open [Compute](https://brev.nvidia.com/org/environments)
3. Choose **Remove** from the compute's menu
4. Enter the compute name and click **Remove Node**