Hardware and Software Requirements for NeMo Platform

This page lists the requirements for the OSS local-install path and self-managed Kubernetes deployments. For the full compatibility table, see the Support Matrix.

For local setup, use the Python package and nemo setup. For self-managed clusters, use the NeMo Platform Helm chart on Kubernetes.

Local Setup Requirements

Component	Requirement	Notes
Python	3.12 or 3.13	Use an isolated virtual environment.
Package installer	`uv>=0.9.14,<0.10.0` recommended; `pip` supported	The one-line installer sets up `uv` automatically when needed.
Operating system	Recent Linux or macOS release	See the Support Matrix for the supported OS list.
Memory	8 GB RAM minimum	16 GB or more is recommended for larger local workflows.
Disk space	16 GB free disk space minimum	Additional space is needed for datasets, job outputs, and local model artifacts.
Network	Outbound HTTPS access	Required for package installation and hosted model-provider APIs.
Model provider	NVIDIA Build, OpenAI, Anthropic, Google Gemini, Ollama, or a custom OpenAI-compatible endpoint	`nemo setup` configures one provider as part of the setup flow.
Browser	Current Chrome, Edge, Firefox, or Safari	Required for the Studio UI.

Self-Managed Kubernetes Requirements

Component	Requirement	Notes
Kubernetes	minikube, kind, EKS, AKS, GKE, OKE, OpenShift, or on-prem Kubernetes	Use a cluster with storage, networking, and image-pull access configured before installing the chart.
Helm	Helm 3	Required to install and upgrade the NeMo Platform chart.
Storage	ReadWriteMany-capable persistent storage	Required for shared files and job storage in multi-pod deployments.
Image access	NGC credentials and Kubernetes image pull secrets	Required to pull NeMo Platform container images.

GPU Requirements

Local provider workflows do not require a local GPU. GPU requirements apply only when you run local model or GPU-accelerated workloads.

Component	Requirement	Notes
GPU	NVIDIA data center GPU with 40 GB VRAM minimum; 80 GB recommended	A100 80GB, H100 80GB, and B200 180GB meet the recommended profile.
CUDA	CUDA 13 or later	GPU workloads require CUDA 13 or later.
NVIDIA driver	CUDA 13-capable driver, R580 branch or later	See the CUDA 13.0 release notes for exact driver minimums by operating system.
Platform	Linux x86_64	Local NVIDIA GPU workloads are not supported on macOS or Windows.

Verify Your Environment

$ python --version
$ nemo --help
$ curl -s http://localhost:8080/health/ready

For setup instructions, see Setup.