OSS
NVIDIA Ships Cosmos3-Super, a 64B World Model Coupling Action Trajectories With Video+Audio
NVIDIA released Cosmos3-Super, a 64B physical-AI world model split into a 32B reasoner plus a 32B generator that couples action trajectories with synchronized video and audio generation, aimed at robotic policy development and synthetic data. It natively understands and produces text, images, video, ambient sound, and physical actions, ranking first among open-weight options on several physical benchmarks. For builders in robotics and simulation, it's a notable open foundation for sim-to-real data generation rather than another chat model.
Source
↳ Follow the thread