Skip to content

Soong CLI

Effortless GPU instance management for Lambda Labs

Soong is a powerful command-line tool that simplifies managing Lambda Labs GPU instances. Launch, monitor, and connect to GPU instances with a single command—no manual configuration required.

Soong CLI

Key Features

🚀 One-Command Instance Management

Launch GPU instances with your preferred model and instance type in seconds. No need to navigate web consoles or remember complex API calls.

soong start --model deepseek-ai/DeepSeek-R1 --instance-type gpu_1x_h100_pcie

🔒 Automated SSH Tunneling

Securely access your services through automatically configured SSH tunnels:

  • Port 8000: SGLang inference server
  • Port 5678: n8n workflow automation
  • Port 8080: Instance status daemon

💰 Built-in Cost Controls

Set maximum runtime limits to prevent unexpected charges. Your instance automatically stops when the time limit is reached.

soong start --max-hours 2  # Auto-stop after 2 hours

🤖 Smart Model Selection

Browse available models and get instant recommendations for the right GPU instance type based on VRAM requirements.

soong models --recommend deepseek-ai/DeepSeek-R1
# Recommends: gpu_1x_h100_pcie (80GB VRAM)

📊 Real-time Monitoring

Check instance status, uptime, and cost at any time:

soong status
# Instance: i-abc123def456 (gpu_1x_h100_pcie)
# Status: running
# Uptime: 1h 23m
# Cost: $2.46

Quick Example

# Configure your Lambda Labs credentials (one-time setup)
soong configure

# Start an instance with DeepSeek-R1
soong start --model deepseek-ai/DeepSeek-R1

# SSH into your instance
soong ssh

# Check status and cost
soong status

# Stop the instance when done
soong stop

What's Next?

Ready to get started? Follow our step-by-step guide:

  • Getting Started

    Set up your Lambda Labs account, install the CLI, and launch your first GPU instance in 5 minutes.

  • Configuration Reference

    Detailed explanation of all configuration options and advanced settings.

  • Command Reference

    Complete documentation for every CLI command with examples.

  • Architecture

    System design, cost controls, and how everything works together.

Why Soong CLI?

Traditional GPU instance management involves:

  • Navigating multiple web consoles
  • Manually configuring SSH keys and tunnels
  • Tracking instance costs in spreadsheets
  • Remembering which instance types support which models

Soong CLI eliminates all of this. Configure once, then launch and manage instances with simple commands. Focus on your work, not infrastructure management.

System Requirements

  • Python: 3.10 or later
  • Operating System: Linux, macOS, or Windows (with WSL)
  • Lambda Labs Account: Sign up here

Community & Support

New to Lambda Labs?

Check out our Prerequisites Guide for step-by-step instructions on setting up your Lambda Labs account.