Integrated AI infrastructure stack
Radiant brings together data centers, GPU hardware, network fabrics, storage, managed compute, managed services, and an operational control plane under one platform and contract.
Radiant
Freemium
Radiant is an integrated AI infrastructure platform that finances, builds, and operates data centers, GPU systems, networking, storage, and managed services. It helps AI teams and infrastructure operators provision and run compute through a unified platform and FlightDeck control plane.
什么是 Radiant？
Radiant is an integrated platform for building and operating AI infrastructure. It combines powered data-center sites, NVIDIA GPU systems, networking, storage, managed compute, managed services, and operational software into one system rather than requiring separate providers for each layer.
The platform provides dedicated bare metal, tenant-isolated GPU virtual machines, managed Kubernetes, and managed Slurm. GPU clusters are designed across compute, interconnect, and storage, and can be consumed through multiple service models while remaining part of the same fleet.
FlightDeck is Radiant’s operational plane for viewing and managing capacity, health, topology, maintenance, identity, security, audit, inventory, and lifecycle. It is available through a dashboard, API, CLI, SDK, and Terraform-ready workflows. Radiant also operates the infrastructure lifecycle, including cluster bring-up, burn-in, topology validation, tuning, fault isolation, and recovery.
Radiant 能做什么？
Multiple GPU consumption models
The same physical GPU fleet can be delivered as dedicated bare-metal clusters, tenant-isolated virtual machines, Kubernetes pools, or Slurm queues, allowing capacity to be allocated according to workload and team requirements.
NVIDIA GPU cluster options
Radiant publishes support for Blackwell Ultra, Hopper, and Rubin-generation systems, with cluster designs that account for GPU topology, networking, interconnects, and storage.
FlightDeck fleet operations
FlightDeck gives operators a shared view of delivered, reserved, healthy, and in-use capacity, along with infrastructure health, topology, maintenance events, and resource state.
Topology-aware placement and recovery
Operators can reserve resource blocks against topology and failure boundaries, trace faults to affected workloads and tenants, and cordon, reset, replace, and restore impacted nodes.
Operational interfaces and governance
The platform exposes dashboard, API, CLI, SDK, and Terraform-ready workflows, with controls for identity, roles, service accounts, MFA, OIDC federation, security groups, encryption, key management, audit trails, and policy events.
使用场景
“Run dedicated large-scale GPU clusters”
Organizations needing dedicated capacity can use bare-metal clusters ranging from thousands to tens of thousands of GPUs, with Radiant handling cluster provisioning and lifecycle operations.
“Provide isolated GPU environments to multiple tenants”
Infrastructure teams can deliver tenant-isolated GPU virtual machines with drivers, frameworks, policy, and identity applied during provisioning, while managing shared capacity from the same fleet layer.
“Operate ML platforms with Kubernetes or Slurm”
ML and infrastructure teams can run GPU-native workloads on managed Kubernetes or use managed Slurm queues for scheduler-based training, inference, simulation, or HPC workflows.
“Place workloads around topology and failure boundaries”
Operators can reserve topology-aware resource blocks and place workloads with awareness of cluster topology, switch fabrics, NVLink domains, storage placement, and failure boundaries.
“Investigate and recover infrastructure faults”
Operations teams can monitor node and network health, trace degraded components to affected workloads, and coordinate cordoning, replacement, reset, and restoration through FlightDeck.
常见问题
What infrastructure does Radiant provide?
Radiant describes a stack covering powered data-center sites, NVIDIA GPU systems, networking, storage, dedicated bare metal, GPU virtual machines, managed Kubernetes, managed Slurm, and related operational services.
How can teams consume Radiant GPU capacity?
GPU capacity can be consumed as dedicated bare-metal clusters, tenant-isolated virtual machines, Kubernetes pools, or Slurm queues. Radiant states that these models use the same underlying physical fleet.
What is FlightDeck used for?
FlightDeck is Radiant’s operational plane for observing and operating AI infrastructure. It covers capacity, health, topology, maintenance, identity, security, audit, inventory, and lifecycle, with access through a dashboard, API, CLI, SDK, and Terraform-ready workflows.
Which GPU generations are listed by Radiant?
Radiant lists Blackwell Ultra, Hopper, and Rubin-generation systems. Specific published examples include the GB300 NVL72, H200, and Vera Rubin NVL72 architectures.
Does Radiant publish pricing or plan limits?
No pricing or plan information is provided in the supplied source material. The available product pages direct visitors to contact Radiant or speak with an infrastructure architect.
快速信息
- Category
- AI infrastructure platform
- Primary platform
- Radiant AI Cloud with FlightDeck operations
- Deployment models
- Bare metal, GPU virtual machines, managed Kubernetes, and managed Slurm
- GPU generations listed
- Blackwell Ultra, Hopper, and Rubin
- Operational access
- Dashboard, API, CLI, SDK, and Terraform-ready workflows
- Pricing
- Not published in the supplied sources
Radiant 替代品
Together AI
together.ai
Together AI 是一个支持推理、微调、GPU 集群、沙盒和托管存储的 AI 云平台。
Paperspace
www.paperspace.com
Paperspace is a cloud platform for developing, training, and deploying machine learning applications with managed notebooks, GPU machines, and deployment workflows. It serves ML developers, data scientists, researchers, and teams that need on-demand accelerated computing.
CUDO Compute
www.cudocompute.com
CUDO Compute designs, deploys, and operates dedicated NVIDIA GPU infrastructure for enterprise AI training and inference. It combines power-ready sites, cluster engineering, commissioning, and ongoing operational support for production workloads.
OVHcloud AI & Machine Learning
www.ovhcloud.com
OVHcloud AI & Machine Learning is a Public Cloud portfolio for building, training, deploying, and integrating AI and machine learning models. It supports data scientists, developers, and organisations working with predictive analytics and generative AI applications.
ComfyICU
comfy.icu
ComfyICU is a managed cloud platform for running, sharing, and deploying ComfyUI workflows. It supports visual workflow development, serverless GPU execution, team workspaces, and REST API deployment without requiring users to manage GPU infrastructure.
Salad
salad.com
面向 AI 工作负载的分布式 GPU 云，按用量计费