Open-source NVIDIA Run:ai, a component of NVIDIA DSX OS, available now in Nebius Applications

NVIDIA has open-sourced its full Run:ai platform, a Kubernetes-native AI workload and GPU orchestration system within the NVIDIA DSX AI Factory Platform, and Nebius is offering it through Nebius AI Cloud Applications on managed Kubernetes. Building on the open-source KAI Scheduler, the platform provides fractional GPU allocation, gang scheduling, quota management, workload submission tools, role-based access controls, and utilization monitoring to help multiple teams share GPU clusters efficiently. Its modular components, including KAI Resource Management, the GPU Fractioning Operator, and Workload Map, can also be deployed independently. NVIDIA and Nebius benchmarks indicate that half-GPU allocations retained 77% of full-GPU throughput while supporting 86% of concurrent-user capacity, and shared deployment of chat, reasoning, and embedding models served about three times as many users as dedicated GPU allocations. The platform is designed to replace static partitions or first-come-first-served cluster access with policies that guarantee team capacity, enable temporary borrowing of unused resources, and reclaim them through preemption when needed.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.