Skip to content

Dedicated NVIDIA GPUs, billed by the hour

RTX 3090 and RTX 4090 servers for AI/ML training and inference, 3D rendering and video. Spin one up for a job, shut it down when it's done.

GPUs per server1 – 8
VRAM per RTX 309024 GB
From$8.50/hr
AccessSSH · RDP

GPU Server plans

Billed hourly. Monthly prices shown for comparison.

GPU Server 8SRTX 3090
$8.50/hr
GPUs
8× RTX 3090 (24 GB)
vCPU
48 vCPU
Memory
240 GB
Storage
4 TB
Network
IPv4 + IPv6
Configure →
GPU Server 9SRTX 4090
Custom
GPUs
8× RTX 4090 (24 GB)
vCPU
64 vCPU
Memory
256 GB
Storage
5 TB
Network
IPv4 + IPv6
Talk to Sales

Compare plans

PlanGPUsvCPU / RAM / DiskHourlyMonthly
GPU Server 8S8× RTX 3090 (24 GB)48 / 240 GB / 4 TB$8.50$4,200
GPU Server 9S8× RTX 4090 (24 GB)64 / 256 GB / 5 TBContact salesContact sales

Built for GPU work

AI/ML trainingUp to 8 GPUs in one server for training runs.
InferenceServe models on a single RTX 3090 from $8.50/hr.
RenderingBurst render jobs and pay only for the hours you use.
VideoTranscoding and processing on dedicated GPUs.

Every GPU Server includes

  • Billed hourly
  • Ubuntu or Windows Server
  • IPv4 and IPv6
  • SSH or RDP access
  • Browser console
  • One-click root password reset