NVIDIA Open-Sources OSMO: One YAML Orchestrates Physical AI Training, Simulation, and Robot Testing

What changed

NVIDIA open-sourced OSMO, the Kubernetes-native orchestrator designed to unify AI workflows across training, simulation, and physical robot testing. It lets teams describe tasks with a single YAML file, which automatically routes work to appropriate hardware ranging from powerful GB200 clusters to edge devices like Jetson AGX Thor. The tool moves infrastructure orchestration out of custom code and into declarative configuration. OSMO is now available under an Apache-2.0 license, with the latest version 6.3.1 ready for use.

Why builders should care

Managing AI pipelines that span training models, simulating environments, and testing on physical robots usually involves juggling separate tools and custom scripts. OSMO simplifies this by combining those stages in one workflow definition, streamlining deployment and scaling. Robotics teams gain consistent, repeatable setups that automatically assign compute resources without writing extra infrastructure code. This frees up engineers to focus on model and robot development rather than pipeline plumbing.

The practical takeaway

For robotics and AI teams, OSMO cuts operational friction across multiple compute environments. It lets developers coordinate compute-heavy simulation and training jobs on server-grade clusters, while seamlessly integrating lower-powered edge hardware for real-world robot testing. This means faster iteration cycles, better resource use, and fewer integration errors. Anyone working on physical AI systems can now adopt OSMO to make their workflows more predictable and scalable out of the box.

What to watch next

Watch how the community adopts OSMO for complex robotics projects beyond NVIDIA’s internal labs. The Kubernetes-native approach may push other AI toolmakers to support unified, declarative orchestration spanning cloud clusters and edge devices. Also, look for integration with popular robotics stacks and cloud services that could broaden OSMO’s reach. How well OSMO handles diverse hardware and scales under heavy loads will determine if it becomes a standard in physical AI development.

AI Quick Briefs Editorial Desk