How fast can I launch a GPU instance?

GPU instances launch in 30-60 seconds. Pre-configured templates with PyTorch, TensorFlow, and CUDA are ready immediately. Custom Docker environments may take 1-2 minutes.

What access methods are available?

You can access your GPU pod via SSH, web-based terminal, Jupyter notebooks, or VS Code remote. All methods support full root access and persistent storage.

Can I save my environment?

Yes, you can create custom templates from running pods to save your exact environment configuration, installed packages, and data for future deployments.

Quick Start Guide

Launch GPU Pod in 3 Simple Steps

From zero to running GPU instance in under 60 seconds. No DevOps experience required.

30-60 sec deployment

$5 free credit

Pre-installed frameworks

3 Steps to Launch

Create Your Account

Free Trial

Create your account with $5 free credit to test any of our products — use:

HASHCODE-voltage-665ab4

Email verification only

$5 instant credit

No payment info needed

Full API access

Create AccountTakes 30 seconds

Choose Your GPU Instance

Browse available pods and select based on your needs. Filter by GPU type, price, or region.

RTX 4090

24GB GDDR6X

$0.39/hr

Inference
Fine-tuning
Development

A100 80GB

80GB HBM2e

$3.76/hr

Large models
Training
Research

H100 80GB

80GB HBM3

$6.62/hr

GPT training
Transformer Engine
FP8 precision

Selection Tips:

Region: Choose closest for lowest latency

Uptime: 24h+ for stable instances

Storage: Check disk space for datasets

Network: Higher bandwidth for data transfer

Browse GPU Pods View Pricing

Connect & Start Working

Configure your pod and access it via SSH, Jupyter, or Web Terminal.

Pod Configuration

Pod Name *

Choose Your Pods

Template * 🐳

SSH Key *

Additional Options

Encrypt Volume (recommended for sensitive data)SSH Terminal AccessStart Jupyter Notebook

Web Terminal Available

Access your pods directly from https://voltagegpu.com/your-pods for active pods

Connection Methods

SSH Access

ssh root@your-pod-ip -p 22

Jupyter Lab

http://your-pod-ip:8888

Template Library

Deploy pre-configured GPU environments instantly. Choose from our collection of optimized templates for AI, ML, and compute workloads.

Browse 100+ Community Templates

Docker Credentials

Secure Docker registry authentication for private container deployments.

Use access tokens for better security

Manage Docker Credentials

SSH Keys

Secure SSH access management for your GPU pods and containers.

Paste your public key content here

Manage SSH Keys

Pre-installed Software

CUDA 11.8 / 12.0

PyTorch 2.0+

TensorFlow 2.13+

JAX / Flax

Transformers

Docker

Jupyter Lab

Ubuntu 22.04

Useful Commands

Check GPU Status

nvidia-smi

Test PyTorch GPU

python -c "import torch; print(torch.cuda.is_available())"

Start Jupyter

jupyter lab --ip=0.0.0.0 --allow-root

Install Package

pip install package-name

Quick Tips for Success

⚡ Launch Time

GPU pods typically launch in 30-60 seconds. High-demand GPUs like H100 may take up to 2 minutes during peak times.

📦 What's Included

Each pod comes with Ubuntu 22.04, CUDA drivers, PyTorch, TensorFlow, JAX, and common ML libraries pre-installed. You also get persistent storage and root access.

💰 Cost Optimization

You can stop your pod anytime to pause billing. Your data remains persistent. Resume within 7 days to keep the same GPU allocation.

🔑 SSH Access

Add SSH keys in Dashboard → SSH Keys before launching, or use the 'Add SSH Key' button in pod details after launch. Keys are automatically deployed to new pods.

💾 Data Persistence

Yes, all pods include persistent NVMe storage. Your data, models, and configurations are preserved even when the pod is stopped.

🐳 Custom Images

You can deploy custom Docker images. Add Docker credentials in your dashboard and select your image when launching.

Ready to Launch Your First GPU Pod?

Get started with $5 free credit. No credit card required.

Start Free Trial Browse Available GPUs

✓ 30-second deployment✓ Pre-configured ML stack✓ 24/7 support