Cosmos platform expands
NVIDIA added new training tools, simulation checkpoints and Cosmos Data Factory blueprints to its Cosmos platform, which helps physical AI systems understand and simulate the real world through world models. The company says it ranks in the top five on six different open model leaderboards.
Data Factory
Cosmos Data Factory brings together tools for labeling, curating, augmenting and evaluating data at scale. The GPU-accelerated Cosmos Curator now supports multimodal and multisensory inputs. Five cloud-native blueprints are available through GitHub, Microsoft Azure, Nebius and CoreWeave: autonomous vehicle dataset generation, image attribute augmentation, event video generation, defect image generation and video data augmentation.
Who is using it
Porsche is leading a shared autonomous vehicle data flywheel across Volkswagen Group brands, using Cosmos Data Factory through Voxel51 on Microsoft Azure. The company expects this to reduce repeated physical test drives by up to 90% and shorten development cycles by up to 85%. ANYbotics generated more than 1,500 synthetic video clips to train its industrial-inspection robots. Ultralytics retrained its YOLO model with real images and targeted synthetic defects.
Training and deployment
At GTC Berlin, NVIDIA extended Cosmos Framework training, optimization and deployment tools as agent skills to more open models and Cosmos 3. Developers can adapt open models such as GR00T, Alpamayo, QWEN, Pi0.5 and compatible Hugging Face models through fine-tuning and reinforcement learning, then optimize with distillation and quantization.
