Overview

NVIDIA NIM is a set of GPU-accelerated inference microservices for pretrained and customized AI models. This appliance packages NVIDIA NIM into an OpenNebula service appliance, enabling OpenNebula to provision virtual machines that run an NVIDIA NIM container on GPU-enabled infrastructure.

The appliance provides a way to deploy an NVIDIA NIM container in an OpenNebula cloud using deployment-time contextualization parameters for registry access and image selection.

Download

The latest version of the NVIDIA NIM appliance can be downloaded from the OpenNebula public Marketplace:

• OpenNebula Marketplace

Requirements

• OpenNebula version: >= 7.2.
• OneGate enabled.
• GPU-enabled infrastructure. Requirements vary depending on the model to be deployed, so check them before deployment.

Release Notes

Details for each release are available on the OpenNebula release page, providing comprehensive information for every version. The NVIDIA NIM appliance is based on Ubuntu 24.04 LTS (x86-64).

Next: NIM Quick Start