NVIDIA open-sourced the control plane and operational components of Kubernetes-based Run:ai. It manages GPU quotas, fair sharing, preemption and fractional use across teams, with UI, CLI and API interfaces. The official announcement says four components, extending beyond the existing KAI scheduler into resource management and workload mapping, can be adopted individually. Free use does not include enterprise support, and NGC distributions have separate terms. Repository publication therefore does not make all installation and support conditions identical.