GPU Eater

World's first cloud service with AMD Radeon

Powerful GPU cloud
for Deep Learning.

Starting at $0.0992 per hour.
Up to 80% cheaper than AWS.


Low cost GPU cloud.
Friendly UI and high-performance.

World's first cloud service with AMD Radeon. TensorFlow pre-installed.
NVIDIA Quadro GPUs are available. Low cost instance for prediction and reasoning.
Persistence container technology enables lightweight operation.
Pay-per-use in seconds rather than hours or months. Fees will be paid by credit card in the next month.

Our clients


Pay per second instead of per hour.

We charge per second, but we have listed hourly and monthly charges here for clarity.
Usage fees include charges for instances, disks, and networks.


Pay per second

  • Pre-set up Machine Learning libraries
  • Pre-configured instances
  • Credit card payment
  • Email support


Charge every month

Pro Customize

Contact us for more detail

  • Full customization plan including the latest generation of graphics card.
    Up to 4 identical cards can be loaded.
  • Pre-set up Deep Learning libraries
  • Full customized instances
  • Credit card payment or bank transfer (USD / JPY)
  • Email and phone support

( Subscription and on-demand )

Subscription plan On-demand plan
On-demand instances
Customized instances
Instances with storage
Using original data
Driver upate
Customer support(E-mail)
Customer support(Phone)
Credit card payment
Bank transfer (USD / JPY)
Long-term discount

Max 80% lower price than AWS.
Optimal price-performance

Provider Plan TFlops Cores GPU Mem Architecture Hourly Monthly
p2.xlarge 4.37 2496 12GB NVIDIA Kepler $0.9000/h $648/m
g3.4xlarge 4.82 2048 8GB NVIDIA Maxwell $1.1400/h $820/m
a1.rx580 6.1 2304 8GB AMD RADEON RX $0.3458/h $249/m
a1.vega56 10.5 3584 8GB AMD RADEON VEGA $0.4794/h $345/m
a1.vegafe 13.1 4096 16GB AMD RADEON VEGA $0.6164/h $443/m
p3.2xlarge 14.0 5120 16GB NVIDIA Volta $3.0600/h $2203/m

* AWS bills additional cost for disk volume usage, IOPS and network traffic usage.
* GPU EATER includes such disk volume and IOPS usage fee.


"1-Click launch" and ultimately efficient UI/UX.
Hyper-quick launch and stop.

With high-performance AMD GPUs,
more Deep Learning opportunities.

High performance,
but low price
comparing to others.
Will be installed in the world's fastest supercomputer by Oak Ridge National Laboratory.
Supports common Machine Learning library such as TensorFlow, Keras, PyTorch.

Use cases

Machine learning applications like Deep Learning, computational fluid dynamics, video encoding, 3D graphics workstation, 3D rendering, VFX, computational finance, seismic analysis, molecular modeling, genomics, and other server-side GPU computation workloads.

In graduate school, I researched robotics and the implementation of image recognition. However, once I finished school my opportunities to program lessened. But even so, my passion for programming stuck with me, and now when I have free time I do experimental projects at home using GPUEATER along with GAN video production system. When I started with deep learning the cost was a huge barrier for me. But, I happened to come across GPUEATER and that let me get my start. For me it's a service that gives you the support you need when you're about to embark on a new challenge.
Mr. Nozawa private customer
Our research group at Quazar Technologies is researching how to use Electrons in two-dimensional materials in devices. These studies are highly computationally intensive, requiring the solution of several million variables for a single calculation. Such calculations have typically been restricted to academic supercomputers. However, our testing has revealed that these can also be efficiently performed on the AMD GPU nodes at We are therefore excited to use commercial cloud computing for cost-effective scientific computation. In addition, we think cloud computing can enable new and efficient computing workflows that can maximize scientific productivity.
Mr.Mani Group Leader, Research Division
In machine learning, the only options are to purchase an expensive GPU or to make use of a GPU instance, and GPUs made by NVIDIA hold the majority of the market share. However, a new option has been proposed by GPUEATER. GPUEATER provides NVIDIA Cloud for inference and AMD GPU clouds for machine learning. I'm sure that you will find an affordable service that meets your needs.
Mr. Negishi private customer



We offer a free trial for businesses,
so please contactinfo@pegara.comafter making an account.