Skip to content
GPU RENTAL / PREPAID / USD

GPU COMPUTE FOR AI

Run your AI on GPUs. An hour, a day or a month.

Choose a server and prepaid period. Check the full cost before launch.

  • Prepay your chosen period
  • Upfront pricing
  • SSH access

FIND YOUR CONFIGURATION

GPU rental & pricing

Configurations: 73 · USD per configuration per hour

A4000 x1 (Σ 16 GB)

A4000 x1 (Σ 16 GB)
$0.20/hour

Daily0.18/h$4.80$4.32−10%0.18/h · prepaid Monthly0.17/h$144.00$122.40−15%0.17/h · prepaid

VRAM / GPU
16 GB
vCPU
4 vCPU
RAM
21 GB RAM
Storage
100 GB storage
Region
NO, Oslo

A6000 x1 (Σ 48 GB)

A6000 x1 (Σ 48 GB)
$0.74/hour

Daily≈ 0.6659/h$17.76$15.98−10%≈ 0.6659/h · prepaid Monthly0.629/h$532.80$452.88−15%0.629/h · prepaid

VRAM / GPU
48 GB
vCPU
6 vCPU
RAM
48 GB RAM
Storage
256 GB storage
Region
US, Kansas City, KS · US, Des Moines, IA

A6000 x1 (Σ 48 GB)

A6000 x1 (Σ 48 GB)
$0.74/hour

Daily≈ 0.6659/h$17.76$15.98−10%≈ 0.6659/h · prepaid Monthly0.629/h$532.80$452.88−15%0.629/h · prepaid

VRAM / GPU
48 GB
vCPU
12 vCPU
RAM
64 GB RAM
Storage
256 GB storage
Region
US, Des Moines, IA · US, Wichita, KS · US, Kansas City, KS

A6000 x1 (Σ 48 GB)

A6000 x1 (Σ 48 GB)
$0.74/hour

Daily≈ 0.6659/h$17.76$15.98−10%≈ 0.6659/h · prepaid Monthly0.629/h$532.80$452.88−15%0.629/h · prepaid

VRAM / GPU
48 GB
vCPU
6 vCPU
RAM
24 GB RAM
Storage
256 GB storage
Region
US, Des Moines, IA · US, Kansas City, KS

RTX5090 x1 (Σ 32 GB)

RTX5090 x1 (Σ 32 GB)
$0.85/hour

Daily0.765/h$20.40$18.36−10%0.765/h · prepaid Monthly0.7225/h$612.00$520.20−15%0.7225/h · prepaid

VRAM / GPU
32 GB
vCPU
12 vCPU
RAM
120 GB RAM
Storage
900 GB storage
Region
NO, Central

A4000 x1 (Σ 16 GB)

A4000 x1 (Σ 16 GB)
$1.04/hour

Daily≈ 0.9359/h$24.96$22.46−10%≈ 0.9359/h · prepaid Monthly0.884/h$748.80$636.48−15%0.884/h · prepaid

VRAM / GPU
16 GB
vCPU
8 vCPU
RAM
45 GB RAM
Storage
250 GB storage
Region
US, New York, NY

L40 x1 (Σ 48 GB)

L40 x1 (Σ 48 GB)
$1.14/hour

Daily≈ 1.0259/h$27.36$24.62−10%≈ 1.0259/h · prepaid Monthly0.969/h$820.80$697.68−15%0.969/h · prepaid

VRAM / GPU
48 GB
vCPU
14 vCPU
RAM
72 GB RAM
Storage
625 GB storage
Region
US, Des Moines, IA

L40S x1 (Σ 48 GB)

L40S x1 (Σ 48 GB)
$1.14/hour

Daily≈ 1.0259/h$27.36$24.62−10%≈ 1.0259/h · prepaid Monthly0.969/h$820.80$697.68−15%0.969/h · prepaid

VRAM / GPU
48 GB
vCPU
12 vCPU
RAM
72 GB RAM
Storage
625 GB storage
Region
US, Des Moines, IA · US, Kansas City, KS

L4 x1 (Σ 24 GB)

L4 x1 (Σ 24 GB)
$1.24/hour

Daily≈ 1.1159/h$29.76$26.78−10%≈ 1.1159/h · prepaid Monthly1.054/h$892.80$758.88−15%1.054/h · prepaid

VRAM / GPU
24 GB
vCPU
8 vCPU
RAM
48 GB RAM
Storage
500 GB storage
Region
PL, Warsaw

RTX6000Ada x1 (Σ 48 GB)

RTX6000Ada x1 (Σ 48 GB)
$1.26/hour

Daily≈ 1.1342/h$30.24$27.22−10%≈ 1.1342/h · prepaid Monthly1.071/h$907.20$771.12−15%1.071/h · prepaid

VRAM / GPU
48 GB
vCPU
12 vCPU
RAM
80 GB RAM
Storage
350 GB storage
Region
US, Kansas City, KS

L40 x1 (Σ 48 GB)

L40 x1 (Σ 48 GB)
$1.30/hour

Daily1.17/h$31.20$28.08−10%1.17/h · prepaid Monthly1.105/h$936.00$795.60−15%1.105/h · prepaid

VRAM / GPU
48 GB
vCPU
28 vCPU
RAM
58 GB RAM
Storage
825 GB storage
Region
CA, Montreal

A6000 x2 (Σ 96 GB)

A6000 x2 (Σ 96 GB)
$1.48/hour

Daily≈ 1.3321/h$35.52$31.97−10%≈ 1.3321/h · prepaid Monthly1.258/h$1,065.60$905.76−15%1.258/h · prepaid

VRAM / GPU
48 GB × 2
vCPU
14 vCPU
RAM
96 GB RAM
Storage
512 GB storage
Region
US, Kansas City, KS · US, Des Moines, IA

FROM SELECTION TO LAUNCH

What will your run cost?

Choose your GPU, set up the environment and estimate your budget. All here, step by step.

The full selected-period price is reserved. It is charged only after actual launch is confirmed. Reviewing price and balance does not launch an instance: final confirmation is separate.

Configure & estimate
  1. 01 / Configure
  2. 02 / Settings
  3. 03 / Estimate

Choose a GPU to configure your server and estimate its cost.

YOUR WORKLOAD → YOUR GPU

Compute for your next idea

Start with your workload. Match memory and the environment to your model and data.

WORKLOAD / GPU

LLM & inference

Run language models and test inference. Memory needs depend on model size, quantization, context length and concurrent requests.

What to consider

Check that model weights and the KV cache fit in VRAM. Leave room for the runtime.

Find a configuration

FROM IDEA TO INSTANCE

Four steps. One dashboard.

Configuration, balance and active instances in one place. No hardware purchase needed.

  1. 01

    Choose your GPU

    Compare configurations, memory and regions. Choose a prepaid hour, day or 30-day month.

  2. 02

    Fund your balance

    Your available balance must cover the full selected period. A new volume created with the server also reserves its first storage hour, charged after server launch is confirmed.

  3. 03

    Connect via SSH

    Choose an available image, create your instance and get connection details in the dashboard.

  4. 04

    Track the expiry time

    Extend before expiry or enable auto-renewal with sufficient balance. Without renewal, the instance is deleted, so keep backups.

CLEAR FROM THE START

More control over your run

Know the price first

Check the full period price and applied discount. The amount is reserved and charged only after launch is confirmed. The hourly equivalent helps compare offers.

A period to fit your task

Choose 1 hour, a day (24 hours) or a month (30 days, 720 hours). Unused prepaid time is not refunded if you delete a running instance early.

Familiar SSH access

Connect to your instance, configure your environment and run your code. Available images depend on the configuration.

FAQ

Before your first run

The essentials on pricing, access and your data.

How does billing work?

For a new instance, choose an hour, a day (24 hours) or a month (30 days, 720 hours). The full period price is reserved in your balance. Charging and paid time begin only after actual launch is confirmed. If the launch is automatically cancelled because the waiting deadline expires, the hold is released without a charge. Existing hourly instances keep their previous billing model.

What balance do I need to launch?

You need the full selected period amount shown before confirmation. If you create a separate volume, you also need its first storage hour. Volumes are billed separately, without the instance term discount.

How are discounts applied?

The selected term discount and final amount are shown before launch. If a global discount is active, the larger of the global and term discounts applies. They do not stack and do not apply to separate volumes.

Can I end a rental early?

Yes, you can delete the instance before expiry. If you delete a running instance early, unused prepaid time is not refunded. Failed launches are handled separately.

Which GPUs and regions are available?

Availability changes. Check current configurations, VRAM, regions and prices in the catalog or dashboard. A specific model is not guaranteed until availability is checked at launch.

How do I connect to my instance?

After creating your instance, open it in the dashboard and use the provided SSH connection details. Choose an image and environment from the options available for your configuration.

What happens to data after deletion?

Deleting an instance, including at expiry without renewal, erases its local disk data. Separate persistent volumes remain and are billed hourly until deleted separately. Keep backups of important data.

How do I renew a rental?

Extend manually before expiry or enable auto-renewal. Auto-renewal requires enough balance for the whole next period. It defaults to on for new hourly, daily and monthly rentals; you can turn it off. Each next period is paid in full before it starts. Without renewal, the instance is deleted when the paid period ends.

YOUR NEXT STEP

Your idea. The compute it needs.

Choose a configuration and check the terms before launch.

Estimate cost

Business invoice request

We will prepare an invoice based on your request. Submitting this form does not create a payment, add funds or launch a server.

Request details

Details are copied from your form. Edit them here for this invoice request; the original form will remain unchanged.

Promo codes do not apply to payments by bank transfer.

By clicking “Send request”, you send these contact and billing details and the selected configuration to the XGPU administrator at support@xgpu.store to prepare your invoice. Do not include passwords, payment card details or secret keys.

Privacy policy