NVIDIA Metropolis

What it does

NVIDIA Metropolis is a vision AI application platform and partner ecosystem for building, deploying, and scaling visual AI agents from edge devices to the cloud. It takes live or archived video and sensor data from cameras and other sensors, and produces analytics, tracking results, and answers to natural language queries about physical spaces such as factories, warehouses, retail stores, airports, and traffic intersections.

Metropolis works as a set of building blocks: video ingestion, storage, and model tuning microservices, plus vision language and computer vision models. Developers can customize vision foundation models with NVIDIA TAO and deploy them with NVIDIA NIM microservices. Described uses include multi-camera tracking, automated visual inspection for defects, intelligent transportation, industrial automation with digital twins through the Mega Omniverse Blueprint, intelligent retail, and robot safety.

Metropolis suits developers, system integrators, and operations teams building vision AI for physical environments. It is not an end-user app. Access is through the developer site at developer.nvidia.com/metropolis. Edge deployments such as robot safety run on NVIDIA IGX hardware, so plan for compatible hardware. Request a quote from NVIDIA for pricing and licensing terms.

What it can do

• Multi-camera tracking across factories, warehouses, retail stores, airports, and traffic intersections
• Natural language queries over live or archived video data
• Automated visual inspection and defect classification with customizable vision AI models
• Video ingestion, storage, and model tuning microservices for building visual AI agents
• Customization of vision foundation models with NVIDIA TAO and deployment with NVIDIA NIM microservices
• Robot safety using on-robot and sensor-infrastructure perception on the NVIDIA IGX platform

Works with

• NVIDIA Omniverse Blueprint (Mega) for digital twin and robot fleet management
• NVIDIA IGX platform for robot safety deployments
• NVIDIA TAO for model customization
• NVIDIA NIM microservices for deployment

Example workflows

Multi-camera search over video archives

Input: live or archived video from multiple cameras in a store or transit hub. Steps: ingest the video with Metropolis ingestion and storage microservices, apply multi-camera tracking, then query in natural language. Output: tracked results and insights for operations teams to act on.

Automated visual defect inspection

Input: images or video of production items. Steps: customize NVIDIA vision AI models for defect classification, deploy them as microservices, and run them on the production line. Output: flagged visual defects to support quality control.

Before you choose

• Request a quote from NVIDIA to confirm pricing and licensing before adoption.
• Robot safety and edge deployments depend on specific NVIDIA hardware such as IGX, so check hardware requirements for your environment.
• Access is developer-oriented; operators interact with agents built on Metropolis rather than with the platform directly.

Listed under

• Coding
• Sandboxes and infrastructure

Get started

Developer page offers learning resources and a 'Try It Now' path to test vision foundation models.

1. Visit developer.nvidia.com/metropolis to review the platform building blocks and learning resources.
2. Identify your use case, such as multi-camera tracking or visual inspection, and the video ingestion, storage, and model components it needs.
3. Customize vision models with NVIDIA TAO if needed, then deploy them with NVIDIA NIM microservices on your target edge or cloud hardware.

Check the specific Metropolis component’s license and deployment costs, including NVIDIA compute and any commercial support.

We couldn’t find published prices for NVIDIA Metropolis. See their pricing page ↗

Sources for this overview

Product pages, documentation, and pricing checked 28 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Metropolis: Intelligent Vision AI for Smart Infrastructure | NVIDIA ↗www.nvidia.com

Company website

nvidia.com: product, pricing and policy pages

Web search

Confirms the official site

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 28 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.