CLI Reference

The Substrate CLI provides a complete interface for managing GPU compute instances from your terminal. All commands support --help for additional details and --json for machine-readable output.

`substrate auth login`

Authenticate with your Substrate account. Opens a browser window for OAuth authentication and stores credentials locally.

Example

$ substrate auth login
Opening browser for authentication...
✓ Authenticated as user@example.com
  Token stored in ~/.substrate/credentials

`substrate compute create`

Create a new GPU compute instance with the specified resource configuration. Substrate composes hardware to match your exact requirements.

Flag	Type	Description
`--gpu-cores`	int	Number of GPU cores to allocate
`--vram`	int	GPU memory in GB
`--ram`	int	System memory in GB
`--storage`	int	NVMe storage in GB
`--region`	string	Deployment region (default: us-east-1)
`--name`	string	Optional instance name for identification

Example

$ substrate compute create --gpu-cores 4 --vram 24 --ram 64 --storage 100 --name training-node-1
✓ Instance created

  ID:       inst_abc123
  Name:     training-node-1
  Status:   provisioning
  Cores:    4 GPU cores
  VRAM:     24 GB
  RAM:      64 GB
  Storage:  100 GB
  Region:   us-east-1

`substrate compute list`

List all active instances in your account. Displays a formatted table with instance details and current status.

Example

$ substrate compute list
ID             NAME               STATUS     CORES   REGION
inst_abc123    training-node-1    running    4       us-east-1
inst_def456    inference-prod     running    8       eu-west-1
inst_ghi789    experiment-3       stopped    2       us-west-2

`substrate compute get <id>`

Show detailed information about a specific instance, including its full resource configuration, status, and endpoint.

Example

$ substrate compute get inst_abc123
ID:         inst_abc123
Name:       training-node-1
Status:     running
Cores:      4 GPU cores
VRAM:       24 GB
RAM:        64 GB
Storage:    100 GB
Region:     us-east-1
Endpoint:   inst_abc123.compute.onsubstrate.run
Created:    2025-01-15T10:30:00Z
Uptime:     24h 12m 5s

`substrate compute ssh <id>`

Open an SSH session to a running instance. Uses your locally configured SSH keys and establishes a secure tunnel.

Example

$ substrate compute ssh inst_abc123
Connecting to inst_abc123.compute.onsubstrate.run...
Welcome to Substrate Compute (Ubuntu 22.04 LTS)

  Instance:  inst_abc123
  GPU:       4 cores / 24 GB VRAM
  RAM:       64 GB
  Storage:   100 GB

user@inst_abc123:~$

`substrate compute stop <id>`

Stop a running instance. The instance storage is preserved and you will continue to be billed for storage only. Use substrate compute create with the same configuration to start a new instance.

Example

$ substrate compute stop inst_abc123
Stopping instance inst_abc123...
✓ Instance stopped. Storage preserved (100 GB).

`substrate compute delete <id>`

Permanently delete an instance and all associated storage. This action is irreversible. You will be prompted for confirmation unless --force is passed.

Flag	Type	Description
`--force`	bool	Skip confirmation prompt

Example

$ substrate compute delete inst_abc123
Are you sure you want to delete inst_abc123? This cannot be undone. [y/N]: y
✓ Instance inst_abc123 deleted.

`substrate status`

Check the health status of Substrate services. Displays the current operational status of the API, compute provisioning, and networking systems.

Example

$ substrate status
Substrate Service Status

  API              ✓ Operational
  Compute          ✓ Operational
  Networking       ✓ Operational
  Storage          ✓ Operational

All systems operational.