About Getting Started with NIM LLM and VLM#

Use the following resources to set up your environment, pull the container, and run your first model. Before you install, confirm that your environment meets the requirements in Prerequisites.

Try a Hosted Model#

Before deploying a model locally, try a NIM API (such as llama-3.3-70b-instruct) at build.nvidia.com to explore the API and model capabilities.

Choose Your Path#

Select the guide that matches your experience level and goals. The following table maps each audience to typical use cases:

Audience

Use Cases

AI/ML engineers and researchers

Quickly test model pipelines, evaluate performance, and prototype new workflows

Platform operators

Run an initial deployment with minimal setup and validate infrastructure compatibility

Technical evaluators

Explore capabilities before committing to production deployment

Refer to Quickstart if you want a working deployment quickly. Refer to Installation if you are preparing a local deployment.

Open the relevant guide from the following cards:

Prerequisites

Hardware, software, driver, and container runtime requirements for deploying NIM LLM and VLM containers.

Prerequisites
Installation

Instructions for configuring network and authentication, logging into Docker, and pulling the NIM container image.

Installation
Configuration

Configure your local cache and advanced settings to customize your NIM LLM and VLM deployment.

Configuration
Quickstart

Step-by-step instructions for deploying model-specific and model-free NIM containers and running inference.

Quickstart
Support Matrix

Supported models, profiles, and hardware platforms for the latest release.

Support Matrix for NIMs
Release Notes

New features, bug fixes, and known issues for the current release.

Release Notes