We’re excited to introduce the beta version of GPU virtual machines. Powered by NVIDIA’s L40S and H100, they are a more straightforward alternative for workloads leveraging single GPUs.

Here’s what is included with each instance:

• Up to 80GB of VRAM

• 500 GB of local NVMe

• Dual 10 Gbps networking for high-speed data transfer.

• $ 0.81 / GPU / hr  on-demand

All GPU instances are deployed in Dallas, US, with Ubuntu 24.04.

Deploy a VM →