NVIDIA has expanded its Cosmos platform with new training tools, simulation checkpoints and Cosmos Data Factory blueprints, aimed at developers building world models, which help physical AI systems understand and simulate the real world. NVIDIA says the platform ranks in the top five on six different open model leaderboards. It pairs the frontier open models with data, training and optimization tools and recipes so builders can scale training data, adapt models, test behavior and optimize performance before deployment.

Better training data

Building world models starts with multimodal, multisensory data. NVIDIA Cosmos Data Factory brings together the tools and skills for labeling, curating, augmenting and evaluating data at scale for training physical AI. Cosmos Curator, a GPU-accelerated solution for curating video training datasets, now supports multimodal multisensory inputs and processing blocks.

Cosmos Data Factory now offers five cloud-native blueprints:

• Autonomous Vehicle Dataset Generation: Curates and labels driving data, then adds synthetic examples of hard-to-capture scenarios to expand training coverage.
• Image Attribute Augmentation: Varies object appearance to improve image search.
• Event Video Generation: Turns still images into synthetic videos of rare or unsafe events.
• Defect Image Generation: Generates printed circuit board, metal and glass defect images for inspection models.
• Video Data Augmentation: Changes weather and lighting in existing footage while preserving scene structure.

Companies are already using these blueprints. Porsche is leading the development of a shared AV data flywheel across Volkswagen Group brands, using Cosmos Data Factory through Voxel51 on Microsoft Azure. The effort is expected to reduce repeated physical test drives by up to 90% and shorten development cycle times by up to 85%.

ANYbotics generated more than 1,500 synthetic video clips with Cosmos Data Factory blueprints. It curated real-world data and validated outputs with Encord's platform, training its industrial-inspection robots to recognize liquid spills and other changes between facility visits. Ultralytics retrained its YOLO model with real images and targeted synthetic defects generated with Cosmos Data Factory, improving defect localization and recall.

Train, adapt and optimize models

At GTC Berlin, NVIDIA extended Cosmos Framework training, optimization and deployment tools and recipes as agent skills to more open models, alongside Cosmos 3. Developers and agents can now adapt open models such as GR00T, Alpamayo, QWEN, Pi0.5 and other compatible Hugging Face models through fine-tuning and reinforcement learning. They can optimize models with distillation and quantization, then deploy with high-performing inference.

Robotics research lab MicroAGI post-trained Cosmos 3 on videos of humans manipulating objects to build a world model for egocentric manipulation. It distilled the 32-billion-parameter reasoner down to an 8-billion-parameter Nano model with no loss in accuracy. Developers can also label data and post-train Cosmos 3 on Roboflow for their specific environments and tasks, then deploy to Roboflow-hosted or on-premises NVIDIA GPUs to bring video reasoning into factories and warehouses.

Test policies and run models faster

New Cosmos 3 Nano model variants for simulation let developers test autonomous policies in repeatable, hard-to-capture, multiview multisensor scenarios. Accompanying code lets developers adapt the models to train and evaluate their own policies for autonomous vehicles and bimanual robots in interactive simulations before real-world testing.

Developers can deploy core Cosmos models using updated Cosmos 3 NIM microservices. These deliver up to 2x faster inference with no quality loss in validated configurations, and continuous video analysis enables sub-second alerts.

For vision AI developers, NVIDIA Metropolis Blueprint for video search and summarization (VSS) 3.3, built on Cosmos NIM microservices, turns one prompt into a video AI application. It helps teams monitor 40% more video streams with the same GPU.

ADNOC uses Fogsphere's MistIQ agentic platform, built with Cosmos and the Metropolis VSS Blueprint, to detect safety hazards in real time across 65 sites. For rare events, Fogsphere also uses Cosmos Data Factory to generate synthetic training data, improving detection without physically staging dangerous situations.

Key Features at a Glance

• Cosmos Data Factory blueprints: Five cloud-native blueprints for curation, labeling, augmentation and synthetic data generation – they cover driving, image, defect and video data needs.
• Porsche AV data flywheel: Up to 90% fewer repeated physical test drives and up to 85% shorter development cycles, as expected by Porsche – less real-world driving needed before deployment.
• Model adaptation: Fine-tuning and reinforcement learning for GR00T, Alpamayo, QWEN, Pi0.5 and compatible Hugging Face models – lets teams specialize open models for their own tasks.
• Distillation example: A 32-billion-parameter reasoner distilled to an 8-billion-parameter Nano model with no loss in accuracy (MicroAGI) – smaller models can cut deployment cost.
• Cosmos 3 NIM microservices: Up to 2x faster inference with no quality loss in validated configurations – faster responses for deployed models.
• Metropolis VSS 3.3: Monitors 40% more video streams with the same GPU – more coverage without extra hardware.
• Cosmos 3 Nano simulation variants: Test autonomous policies in repeatable, multiview multisensor scenarios before real-world testing – safer, cheaper policy validation.

Availability

Cosmos Data Factory blueprints are available through GitHub, Microsoft Azure, Nebius and CoreWeave. The article does not give a specific release date or pricing for these updates.