NVIDIA Announces Major Release of Cosmos World Foundation Models and Physical AI Data Tools
NVIDIA has announced a major release of NVIDIA Cosmos™ world foundation models (WFMs), introducing new tools for physical AI development. The release includes an open reasoning model and enhanced world generation control capabilities.
Key components include:
- Cosmos Transfer WFMs: Converts structured video inputs into controllable photoreal video outputs for synthetic data generation
- Cosmos Predict: Enables multi-frame generation and predicts motion trajectories
- Cosmos Reason: An open, customizable WFM with spatiotemporal awareness for understanding video data
Industry leaders including 1X, Agility Robotics, Figure AI, and Uber are among early adopters. The models are available for preview in the NVIDIA API catalog and listed in the Vertex AI Model Garden on Google Cloud, with Cosmos Predict and Transfer openly available on Hugging Face and GitHub.
NVIDIA ha annunciato un'importante release dei modelli fondazione del mondo NVIDIA Cosmos™ (WFMs), introducendo nuovi strumenti per lo sviluppo dell'IA fisica. La release include un modello di ragionamento aperto e capacità avanzate di controllo della generazione del mondo.
I componenti chiave includono:
- Cosmos Transfer WFMs: Converte input video strutturati in output video fotorealistici controllabili per la generazione di dati sintetici
- Cosmos Predict: Abilita la generazione di più fotogrammi e prevede le traiettorie di movimento
- Cosmos Reason: Un WFM aperto e personalizzabile con consapevolezza spaziotemporale per comprendere i dati video
I leader del settore, tra cui 1X, Agility Robotics, Figure AI e Uber, sono tra i primi adottanti. I modelli sono disponibili per la preview nel catalogo API di NVIDIA e sono elencati nel Vertex AI Model Garden su Google Cloud, con Cosmos Predict e Transfer disponibili apertamente su Hugging Face e GitHub.
NVIDIA ha anunciado un importante lanzamiento de los modelos de fundación del mundo NVIDIA Cosmos™ (WFMs), introduciendo nuevas herramientas para el desarrollo de IA física. El lanzamiento incluye un modelo de razonamiento abierto y capacidades mejoradas de control de generación del mundo.
Los componentes clave incluyen:
- Cosmos Transfer WFMs: Convierte entradas de video estructuradas en salidas de video fotorealistas controlables para la generación de datos sintéticos
- Cosmos Predict: Permite la generación de múltiples fotogramas y predice trayectorias de movimiento
- Cosmos Reason: Un WFM abierto y personalizable con conciencia espaciotemporal para entender los datos de video
Líderes de la industria como 1X, Agility Robotics, Figure AI y Uber están entre los primeros adoptantes. Los modelos están disponibles para vista previa en el catálogo de API de NVIDIA y listados en el Vertex AI Model Garden de Google Cloud, con Cosmos Predict y Transfer disponibles abiertamente en Hugging Face y GitHub.
NVIDIA는 NVIDIA Cosmos™ 세계 기초 모델(WFMs)의 주요 출시를 발표하며 물리적 AI 개발을 위한 새로운 도구를 도입했습니다. 이번 출시에는 개방형 추론 모델과 향상된 세계 생성 제어 기능이 포함되어 있습니다.
주요 구성 요소는 다음과 같습니다:
- Cosmos Transfer WFMs: 구조화된 비디오 입력을 제어 가능한 포토리얼 비디오 출력으로 변환하여 합성 데이터 생성을 가능하게 합니다
- Cosmos Predict: 다중 프레임 생성을 가능하게 하고 운동 궤적을 예측합니다
- Cosmos Reason: 비디오 데이터를 이해하기 위한 시공간 인식을 갖춘 개방형, 맞춤형 WFM입니다
1X, Agility Robotics, Figure AI 및 Uber와 같은 업계 선두주자들이 초기 채택자 중에 포함되어 있습니다. 이 모델들은 NVIDIA API 카탈로그에서 미리 보기로 제공되며, Google Cloud의 Vertex AI Model Garden에 나열되어 있으며, Cosmos Predict와 Transfer는 Hugging Face와 GitHub에서 공개적으로 이용 가능합니다.
NVIDIA a annoncé une publication majeure des modèles de fondation du monde NVIDIA Cosmos™ (WFMs), introduisant de nouveaux outils pour le développement de l'IA physique. La publication comprend un modèle de raisonnement ouvert et des capacités améliorées de contrôle de génération du monde.
Les composants clés incluent:
- Cosmos Transfer WFMs: Convertit des entrées vidéo structurées en sorties vidéo photoréalistes contrôlables pour la génération de données synthétiques
- Cosmos Predict: Permet la génération de plusieurs images et prédit les trajectoires de mouvement
- Cosmos Reason: Un WFM ouvert et personnalisable avec une conscience spatiotemporelle pour comprendre les données vidéo
Des leaders de l'industrie tels que 1X, Agility Robotics, Figure AI et Uber font partie des premiers adoptants. Les modèles sont disponibles en aperçu dans le catalogue API de NVIDIA et répertoriés dans le Vertex AI Model Garden sur Google Cloud, avec Cosmos Predict et Transfer disponibles publiquement sur Hugging Face et GitHub.
NVIDIA hat eine wichtige Veröffentlichung der NVIDIA Cosmos™ Weltgrundlagenmodelle (WFMs) angekündigt und neue Werkzeuge für die Entwicklung physikalischer KI eingeführt. Die Veröffentlichung umfasst ein offenes Denkmodell und verbesserte Steuerungsfunktionen für die Welterzeugung.
Wichtige Komponenten sind:
- Cosmos Transfer WFMs: Wandelt strukturierte Videoeingaben in kontrollierbare fotorealistische Videoausgaben zur Generierung synthetischer Daten um
- Cosmos Predict: Ermöglicht die Generierung mehrerer Frames und sagt Bewegungsbahnen voraus
- Cosmos Reason: Ein offenes, anpassbares WFM mit raum-zeitlichem Bewusstsein zur Analyse von Videodaten
Branchenführer wie 1X, Agility Robotics, Figure AI und Uber gehören zu den frühen Anwendern. Die Modelle sind zur Vorschau im NVIDIA API-Katalog verfügbar und im Vertex AI Model Garden auf Google Cloud gelistet, wobei Cosmos Predict und Transfer offen auf Hugging Face und GitHub verfügbar sind.
- Launch of new AI tools expanding NVIDIA's product portfolio in physical AI
- Partnership with major industry leaders as early adopters
- Wide availability across multiple platforms (API catalog, Google Cloud, Hugging Face, GitHub)
- Integration of responsible AI features through collaboration with Google DeepMind
- Early access limitation for Cosmos Reason component
- Requires specific NVIDIA hardware (Grace Blackwell NVL72 systems) for real-time world generation
Insights
NVIDIA's release of Cosmos world foundation models represents a significant strategic expansion that strengthens the company's AI ecosystem moat. By addressing critical bottlenecks in physical AI development, NVIDIA creates powerful demand drivers for both its software platforms and hardware offerings.
The introduction of three key components – Cosmos Transfer for synthetic data generation, Cosmos Predict for world generation, and Cosmos Reason for multimodal reasoning – positions NVIDIA at the center of the emerging physical AI market. Early adoption by companies like 1X, Agility Robotics, Figure AI, and Uber validates market demand and indicates new revenue opportunities.
This product expansion follows NVIDIA's proven strategy of creating comprehensive platform solutions that drive sustained hardware demand, similar to how CUDA and AI software frameworks have supported GPU sales growth. The specific mention of Grace Blackwell NVL72 systems for inference compute suggests a calculated strategy to drive adoption of NVIDIA's highest-margin products.
By enabling developers to generate massive synthetic datasets through simulation rather than real-world collection, NVIDIA addresses a fundamental cost and time constraint in robotics and autonomous vehicle development. This could accelerate industry adoption rates and expand NVIDIA's total addressable market.
The integration with Google Cloud's Vertex AI Model Garden also expands distribution channels, potentially broadening market reach. While specific monetization details aren't provided, this development reinforces NVIDIA's position as the infrastructure provider of choice for the next wave of AI applications beyond current generative AI use cases.
NVIDIA's Cosmos world foundation models represent a technical breakthrough in bridging the gap between virtual and physical AI applications. The system architecture addresses the fundamental challenge that has constrained physical AI development – the massive data requirements for training robust perception and control systems.
The Cosmos Transfer capability transforms structured simulation data into photorealistic outputs, allowing developers to generate millions of training examples with precise ground truth labels – a process that would be prohibitively expensive and time-consuming with real-world data collection. For robotics companies like Agility and Figure AI, this means being able to simulate rare edge cases and failure modes without physical testing.
The Cosmos Predict models with multi-frame generation capabilities solve the complex problem of trajectory prediction – essential for autonomous systems that must anticipate how objects will move through space. The ability to predict intermediate actions or motion trajectories when given start and end input images is particularly valuable for planning algorithms.
Most technically impressive is Cosmos Reason, which introduces chain-of-thought reasoning to understanding video data and predicting physical interactions. This open and customizable model brings language model capabilities to physical AI, potentially allowing robots to develop more sophisticated understanding of cause and effect.
By making these models available through multiple channels (API catalog, Hugging Face, GitHub) while maintaining integration with NVIDIA's hardware stack, the company has created an elegant technical solution that maintains platform control while fostering developer ecosystem growth. The real-time world generation capabilities specifically optimized for Grace Blackwell systems creates a technical incentive for customers to adopt NVIDIA's full stack approach.
- New Models Enable Prediction, Controllable World Generation and Reasoning for Physical AI
- Two New Blueprints Deliver Massive Physical AI Synthetic Data Generation for Robot and Autonomous Vehicle Post-Training
- 1X, Agility Robotics, Figure AI, Skild AI Among Early Adopters
SAN JOSE, Calif., March 18, 2025 (GLOBE NEWSWIRE) -- GTC—NVIDIA today announced a major release of new NVIDIA Cosmos™ world foundation models (WFMs), introducing an open and fully customizable reasoning model for physical AI development and giving developers unprecedented control over world generation.
NVIDIA is also launching two new blueprints — powered by the NVIDIA Omniverse™ and Cosmos platforms — that provide developers with massive, controllable synthetic data generation engines for post-training robots and autonomous vehicles.
Industry leaders including 1X, Agility Robotics, Figure AI, Foretellix, Skild AI and Uber are among the first to adopt Cosmos to generate richer training data for physical AI faster and at scale.
“Just as large language models revolutionized generative and agentic AI, Cosmos world foundation models are a breakthrough for physical AI,” said Jensen Huang, founder and CEO of NVIDIA. “Cosmos introduces an open and fully customizable reasoning model for physical AI and unlocks opportunities for step-function advances in robotics and the physical industries.”
Cosmos Transfer for Synthetic Data Generation
Cosmos Transfer WFMs ingest structured video inputs such as segmentation maps, depth maps, lidar scans, pose estimation maps and trajectory maps to generate controllable photoreal video outputs.
Cosmos Transfer streamlines perception AI training, transforming 3D simulations or ground truth created in Omniverse into photorealistic videos for large-scale, controllable synthetic data generation.
Agility Robotics will be an early adopter of Cosmos Transfer and Omniverse for large-scale synthetic data generation to train its robot models.
“Cosmos offers us an opportunity to scale our photorealistic training data beyond what we can feasibly collect in the real world,” said Pras Velagapudi, chief technology officer of Agility Robotics. “We’re excited to see what new performance we can unlock with the platform, while making the most use of the physics-based simulation data we already have.”
The NVIDIA Omniverse Blueprint for autonomous vehicle simulation uses Cosmos Transfer to amplify variations of physically based sensor data. With the blueprint, Foretellix can enhance behavioral scenarios by varying conditions like weather and lighting for diverse driving datasets. Parallel Domain is also using the blueprint to apply similar variation to its sensor simulation.
The NVIDIA GR00T Blueprint for synthetic manipulation motion generation combines Omniverse and Cosmos Transfer to generate diverse datasets at scale, benefiting from OpenUSD-powered simulations and reducing data collection and augmentation time from days to hours.
Cosmos Predict for Intelligent World Generation
Announced at the CES trade show in January, Cosmos Predict WFMs generate virtual world states from multimodal inputs like text, images and video. New Cosmos Predict models will enable multi-frame generation, predicting intermediate actions or motion trajectories when given start and end input images. Purpose-built for post-training, these models can be customized using NVIDIA’s openly available physical AI dataset.
With the inference compute power of NVIDIA Grace Blackwell NVL72 systems and their large NVIDIA NVLink™ domain, developers can achieve real-time world generation.
1X is using Cosmos Predict and Cosmos Transfer to train its new humanoid robot NEO Gamma. Robot brain developer Skild AI is tapping into Cosmos Transfer to augment synthetic datasets for its robots. Plus, Nexar and Oxa are using Cosmos Predict to advance their autonomous driving systems.
Multimodal Reasoning for Physical AI
Cosmos Reason is an open, fully customizable WFM with spatiotemporal awareness that uses chain-of-thought reasoning to understand video data and predict the outcomes of interactions — such as a person stepping into a crosswalk or a box falling from a shelf — in natural language.
Developers can use Cosmos Reason to improve physical AI data annotation and curation, enhance existing world foundation models or create new vision language action models. They can also post-train it to build high-level planners to tell the physical AI what it needs to do to complete a task.
Accelerating Data Curation and Post-Training for Physical AI
Based on their downstream task, developers can post-train Cosmos WFMs using native PyTorch scripts or the NVIDIA NeMo™ framework on NVIDIA DGX™ Cloud.
Cosmos developers can also use NVIDIA NeMo Curator on DGX Cloud for accelerated data processing and curation. Linker Vision and Milestone Systems are using it for curating large amounts of video data to train large vision language models for visual agents built on the NVIDIA AI Blueprint for video search and summarization. Virtual Incision is exploring it to be deployed in future surgical robots, while Uber and Waabi are advancing autonomous vehicles development.
Driving Responsible AI and Content Transparency
In line with NVIDIA’s trustworthy AI principles, NVIDIA enforces open guardrails across all Cosmos WFMs. In addition, NVIDIA is collaborating with Google DeepMind to integrate SynthID to watermark and help identify AI-generated outputs from the Cosmos WFM NVIDIA NIM™ microservice featured on build.nvidia.com.
Availability
Cosmos WFMs are available for preview in the NVIDIA API catalog and now listed in the Vertex AI Model Garden on Google Cloud. Cosmos Predict and Cosmos Transfer are openly available on Hugging Face and GitHub. Cosmos Reason is available in early access.
Learn more by watching the NVIDIA GTC keynote and by registering for Cosmos sessions and training from NVIDIA and industry leaders at the show, including “An Introduction to Cosmos World Foundation Models” with Ming-Yu Liu, vice president of generative AI research at NVIDIA.
About NVIDIA
NVIDIA (NASDAQ: NVDA) is the world leader in accelerated computing.
For further information, contact:
Paris Fox
Corporate Communications
NVIDIA Corporation
+1-408-242-0035
pfox@nvidia.com
Certain statements in this press release including, but not limited to, statements as to: the benefits, impact, availability, and performance of NVIDIA’s products, services, and technologies; third parties adopting NVIDIA’s products and technologies and the benefits and impact thereof; and Cosmos opening opportunities for step-function advances in robotics and the physical industries are forward-looking statements that are subject to risks and uncertainties that could cause results to be materially different than expectations. Important factors that could cause actual results to differ materially include: global economic conditions; our reliance on third parties to manufacture, assemble, package and test our products; the impact of technological development and competition; development of new products and technologies or enhancements to our existing product and technologies; market acceptance of our products or our partners' products; design, manufacturing or software defects; changes in consumer preferences or demands; changes in industry standards and interfaces; unexpected loss of performance of our products or technologies when integrated into systems; as well as other factors detailed from time to time in the most recent reports NVIDIA files with the Securities and Exchange Commission, or SEC, including, but not limited to, its annual report on Form 10-K and quarterly reports on Form 10-Q. Copies of reports filed with the SEC are posted on the company's website and are available from NVIDIA without charge. These forward-looking statements are not guarantees of future performance and speak only as of the date hereof, and, except as required by law, NVIDIA disclaims any obligation to update these forward-looking statements to reflect future events or circumstances.
© 2025 NVIDIA Corporation. All rights reserved. NVIDIA, the NVIDIA logo, NVIDIA Cosmos, NVIDIA DGX, NVIDIA NeMo, NVIDIA NIM, NVIDIA Omniverse and NVLink are trademarks and/or registered trademarks of NVIDIA Corporation in the U.S. and other countries. Other company and product names may be trademarks of the respective companies with which they are associated. Features, pricing, availability and specifications are subject to change without notice.
A photo accompanying this announcement is available at https://www.globenewswire.com/NewsRoom/AttachmentNg/6c781321-9544-4bbf-bb47-8bab73fe2f63
FAQ
What are the key features of NVIDIA's new Cosmos world foundation models?
Which companies are early adopters of NVIDIA's Cosmos WFMs?
Where can developers access NVIDIA's new Cosmos models?
How does NVIDIA ensure responsible AI in Cosmos WFMs?
