Private high-speed fabric
H200 nodes use a 3.2 Tb/s topology; A100 nodes use a 1.6 Tb/s topology for distributed collectives.
GPU Clusters
One cluster. 16–64 GPUs moving as one.
Create 2–8 homogeneous nodes directly from the dashboard, each with eight GPUs, a private high-speed fabric, optional network storage and a reusable Pod template.
Transparent baseline
$1.79/GPU/hr
A100 SXM cluster on-demand
2–8
Self-service nodes
16–64
GPUs per cluster
1.6–3.2 Tb/s
Private fabric
EU regions, encrypted storage and network isolation options
Why GPURento
Less infrastructure work. More model progress.
Repeatable environments
Launch every node from the same tested Pod template and mount one region-compatible network volume.
Two orchestrators
Choose PyTorch Distributed or managed Slurm when creating the cluster.
From code to capacity
One clean path to production.
1
Choose topology
Pick accelerator, node count, region, fabric and capacity model.
2
Attach the environment
Choose PyTorch or Slurm, a Pod template and optional shared storage.
3
Create without a sales form
Prepaid resources need a positive wallet balance. The $50 minimum applies to deposits, not to your remaining balance.
deploy.py
await fetch("/api/clusters", {
method: "POST",
headers: { "content-type": "application/json" },
body: JSON.stringify({
name: "training-fabric",
gpuId: "h200-sxm",
region: "eu-west-par",
nodeCount: 2,
scheduler: "slurm",
image: "pytorch/pytorch:2.6.0-cuda12.6-cudnn9-runtime",
templateId: null,
storageVolumeId: null
})
})✓ Configuration saved to the workspace · provisioning state visible
Included
A serious platform, not a thin wrapper.
Every workload gets the same API, identity, billing and observability layer.
2–8 homogeneous GPU nodes
Eight GPUs per node
H200 or A100 cluster pricing
Private high-speed network
Managed Slurm scheduler
PyTorch Distributed mode
Region-scoped network volumes
Reusable Pod templates
When should I use a cluster instead of a single instance?+
Use a cluster when the model, dataset or time-to-result target requires multiple nodes and a fast interconnect. Smaller fine-tuning jobs often remain simpler on one multi-GPU node.
Is Slurm available without contacting sales?+
Configure resources directly in the dashboard. Prepaid usage uses wallet credit; monthly rentals have a separate checkout.
Can I scale the cluster myself?+
Yes. The dashboard supports scaling between 2 and 8 nodes while preserving the GPU type, region and fabric.
Can a cluster use persistent storage?+
Yes. Create a network volume first, then attach it during cluster creation. The volume and cluster must be in the same region.
Ready when you are