Launching Your First Instance

Step-by-Step (Dashboard)

1. Choose Your GPU

Navigate to Instances → Launch New. You’ll see a pricing table with all available GPUs:

Prices are live and reflect the cheapest available provider.

2. Select a Template

Choose a pre-configured environment:

• PyTorch 2.3 — ML training and fine-tuning
• Jupyter Lab — Interactive notebooks
• ComfyUI — Stable Diffusion workflows
• vLLM — LLM inference serving
• Custom — Your own Docker image

3. Configure Options

• Region: US, EU, Asia, or Any (cheapest globally)
• Disk: 50 GB default (adjustable)
• Spot: Toggle for cheaper spot instances (may be interrupted)
• SSH Key: Select from your saved keys

4. Launch

Click Launch Instance. Pulserun will:

1. Find the cheapest available provider
2. Provision your GPU instance
3. Install your selected template
4. Attach your SSH key

Typical provisioning time: 20-60 seconds.

5. Connect

Once running, you’ll see:

• SSH command: ssh root@45.123.67.89
• Jupyter URL: http://45.123.67.89:8888
• Status: Running with live cost counter

6. Managing Your Instance

• Stop — Pause billing, keep disk state
• Start — Resume a stopped instance
• Terminate — Delete instance and disk permanently