Quick to get started picto

Quick to get started

GPU instances work exactly like other instances: they are billed by the second and integrate seamlessly with your existing infrastructure.

Pass-through access picto

Pass-through access

Tap the GPU's full power with direct passthrough access. Speed-up Tensorflow or any other Deep Learning, Big Data, AI, or Encoding workload.

Cloud native GPUs picto

Cloud native GPUs

Start your GPU instances via our API or with your existing deployment management tools. Add parallel computational power to your stack with no effort.

From Artificial Intelligence to Deep Learning, you can now move your HPC workload to the Cloud

Exoscale’s GPU instances are ready to be used with the framework of your choice, would that be Tensorflow, Theano, Caffe2, or PyTorch. Optimize costs and performance, with maximum computational power, on demand at your fingertips.

Use your favorite AI and Machine Learning framework on Exoscale's on-demand GPU instances

GPU2 Instances - Unparalleled Performance for Tensor Operations such as Deep Learning and Neural Networks, Based on NVIDIA Tesla V100 Graphic Cards

With over 100 teraFLOPS of Deep Learning performance, AI models that would consume weeks of computing resources on previous systems can now be trained in a few days.

Compared to the P100, the V100 benefits from overall increased performance: it has nearly the double of single precision and double precision teraflops, and 50% increased memory bandwith. However, where the V100 really shines is by having dedicated cores for tensor operations such as Deep Learning and Neural Networks.


  • From 1 up to 4 NVIDIA Tesla V100 per GPU2 instance
  • 5,120 Cuda cores per card
  • 640 dedicated Tensor Cores per card
  • 16 GB of dedicated RAM per card
  • Up to 48 CPUs per GPU2 instance
  • Up to 1.6 TB of SSD local storage


  • Artificial Intelligence
  • Deep Learning and Machine Learning with Tensorflow, Torch, Caffe
  • Trend forecasting
  • Heavy HPC workloads like quantum chemistry, molecular dynamics, etc.


Tesla V100 GPU in the Cloud

Fully Predictable Pricing, Billed by the Second.

GPU instances based on Tesla V100 come in four instance types spanning from 1 to 4 GPUs. They are priced to the second, and come with fast SSD storage from 100GB up to 1.6TB depending on the instance type. See detailed prices

RAM CPU Cores GPU Cards Local Storage Price ({{ currency | uppercase }})
Small 56 GB 12 Cores 1 GPU min 100 GB max 800 GB {{ prices.opencompute.gpu2.small[currency] | number:5 }} Hour {{ prices.opencompute.gpu2.small[currency] * 720 | number:2 }} Month
Medium 90 GB 16 Cores 2 GPU min 100 GB max 1.2 TB {{ prices.opencompute.gpu2.medium[currency] | number:5 }} Hour {{ prices.opencompute.gpu2.medium[currency] * 720 | number:2 }} Month
Large 120 GB 24 Cores 3 GPU min 100 GB max 1.6 TB {{ prices.opencompute.gpu2.large[currency] | number:5 }} Hour {{ prices.opencompute.gpu2.large[currency] * 720 | number:2 }} Month
Huge 225 GB 48 Cores 4 GPU min 100 GB max 1.6 TB {{ prices.opencompute.gpu2.huge[currency] | number:5 }} Hour {{ prices.opencompute.gpu2.huge[currency] * 720 | number:2 }} Month
  1. [1]

    Local Storage is not included in the displayed Instances price, and has a cost of {{ prices.opencompute.volume[currency] | number:5 }} {{ currency | uppercase }} / GB hour. Learn more about Local Storage

  2. [2]

    Instance needs to be shutdown for hypervisor and platform updates as it cannot be live-migrated

  3. [3]

    Monthly prices are estimated for a 720 hours month (hourly price * 720)

  4. [4]

    All prices are rounded. For a precise and detailed view of our prices have a look at our official pricelist (pdf)

  5. [5]

    Available in the following Zones: AT-VIE-1

  6. [6]

    Please note that GPU instances require account validation: access is provided with priority to established businesses, and is granted after a manual screening process

GPU Cloud Servers Features

Check out GPU pricing and access GPU cloud servers
Choose From a Wide Selection of Officially Supported Templates
Multiple OS images support picto

Trusted by engineers across Europe.

When running mission critical production workloads in the cloud, a partner you can rely on makes all the difference. Our customer success engineers have helped hundreds of customers from all over Europe migrate, run and scale production workloads on Exoscale.

Contact us