NVIDIA releases four open Cosmos 3 Nano models for world simulation
NVIDIA has added four new open-weight models to its Cosmos 3 platform, designed for generating interactive world simulations and multi-sensor driving scenes. The models—Cosmos3-Nano-Transfer-Auto, Cosmos3-Nano-Sim-Auto, Cosmos3-Nano-Sim-Transfer, and Cosmos3-Nano-Sim-Bimanual—each specialize in different tasks like synchronized camera/LiDAR data, photorealistic video from simulator controls, and…
Key points
- Four new open-weight Cosmos 3 Nano models released for interactive world simulation and driving scenarios
- Models include multi-view driving data, synchronized camera/LiDAR streams, photorealistic video, and robot action prediction
- All use OpenMDW1.1 license; require Linux, NVIDIA GPU, Docker, and Container Toolkit for inference
The models aim to address gaps in simulation environments for autonomous driving and robotics. Transfer-Auto generates multi-view driving data with variable weather and lighting, while Sim-Auto streams synchronized video and LiDAR for real-time driving simulations. Sim-Transfer turns physics simulator controls into photorealistic video, and Sim-Bimanual predicts visual outcomes from robot or camera actions. NVIDIA provides technical reports, notebooks, and open skills for inference, though users need Linux, an NVIDIA GPU, Docker, and Container Toolkit to run them. The focus now shifts to evaluating how well these models support policy learning and identifying their limitations.
New NVIDIA Cosmos 3 Open Models for Interactive World Simulation
Hugging Face · 30 October 2026
Loading the full article…
This text was published by Hugging Face. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Research
All →- Researchers link AI agent power to human Goal attainment in formal models · 1 src
- PromptOps framework proposed for iterative LLM judge optimization · 1 src
- Researchers introduce BCSS to scale symbolic models of reasoning · 1 src
- Researchers test Cross-Agent formalization for clinical trial criteria · 1 src
- Researchers introduce efxplorer for Cooperative Multi-Agent reinforcement learning · 1 src
Comments
via GitHub Discussions