Deploy a GPU Server
Hourly cloud GPU VMs. Available with our TensorML 20.04 template
requires 1 command (jupyter notebook --ip=0.0.0.0) to start a Jupyter Notebook!
Fully dedicated resources. Deploy in Chicago, New York, or Las Vegas. Contact support if you encounter any issues — we're here to help!
1. Choose a graphics card model
ML/AI GPUs by NVIDIA
We compare tensor core performance because these cards' primary use case is ML.
A100 PCIE 80GB VRAM
312 TFLOPS TF32: Most powerful AI GPU with more VRAM, in a PCIE format.
V100 SXM 16GB VRAM
125 TFLOPS TF32: Industry standard AI GPU, available NVLink'd together.
V100 PCIE 16GB VRAM
112 TFLOPS TF32: Industry standard AI GPU, great price-to-performance ratio.
General Purpose GPUs by NVIDIA
We compare FP32 performance because these cards can be used for any use case.
2. Select how many GPUs
3. Customize RAM amount
RAM costs $0.005/GB/hour. Check our instance restrictions page for limits. Contact us to add additional options.
4. Customize vCPU count
Each vCPU (1 CPU thread) costs $0.01/vCPU/hour. Check our instance restrictions page for limits. Contact us to add additional options.
5. Customize storage
Contact us to add additional options.
Learn more about our storage classes. Note: deploying Windows 10 VMs on HDD storage is NOT recommended.
6. Select a location
Not all locations will be available for a selected machine configuration. Learn more about our data centers.
Chicago DC1, USA Best Overall Stock
Our largest location with the broadest set of GPUs available. Less than 50ms to all major U.S. cities.
Las Vegas DC1, USA Best VFX GPU Availability
Dozens of racks of A40s and other VFX-optimized GPUs. 10ms to Los Angeles, 35ms to Seattle.
7. Select an operating system
Operating system templates updated on March 10, 2023. Check out our Windows 10 Parsec cloud gaming article if this will apply for you.
TensorML 20 TensorFlow ML / AI
Jupyter, TensorFlow, Keras, CUDA, 30GB
Requires 40GB+ disk
TensorML 20 Everything ML / AI
Jupyter, RAPIDS, TensorFlow, PyTorch, Keras, fastai, CUDA, 53GB
Requires 60GB+ disk
8. Configure details
Your actual charge will be pro-rated to the
1x Undefined GPU
2 GB RAM
20 GB NVMe SSD
Chicago DC1, USA
TensorML 20.04 LTS
Learn more here.
We don't know the exact stock at the moment, but your selected machine configuration should be available. If your server isn't provisioned within 10 minutes, delete it and contact support.
Cost per hour when the VM is running.
Cost per hour when the VM is stopped.
Email us at firstname.lastname@example.org if you
spending over $10k/month and are interested in
a long term contract. Save 25% with a 12 month
commit and 40% with a 24 month commit.
Share Your Server
Share your server's specs using a custom link.
Pressing the button auto-copies the link.
Generate and Copy Link