NVIDIA Cosmos™ is a platform purpose-built for physical AI, featuring state-of-the-art generative world foundation models (WFMs), guardrails, and an accelerated data processing and curation pipeline. Developers use Cosmos to accelerate physical AI development for autonomous vehicles (AVs), robots, and video analytics AI agents.
Developer documentation for Cosmos core models, including installation, quick-start, and post-training guides
A comprehensive guide for adapting Cosmos core models for various deployments—includes data curation, evaluation, and post-training guides using real-world datasets.
User guide and API reference for curating video datasets on DGX Cloud
Deployment guide for NIM microservices for Cosmos world foundation models
Documentation for the Cosmos-Reason1 NIM, which allows for reasoning with a vision language model (VLM).
Documentation for the Cosmos-Embed1 NIM, which generates joint video-text embeddings for short-form videos