RTX PRO 4000 GPU
- Cores / threads
- 14 / 20
- GPU
- 1 × RTX PRO 4000, 24 GB
- Memory
- 64 GB DDR4
- Storage
- 2 × 512 GB NVMe
- Location
- Helsinki, Finland
$329per month
$259 setup
NVIDIA RTX PRO Blackwell GPUs with up to 384 GB of VRAM, for AI inference, training and rendering.
| Server | Cores | Memory | Storage | Location | Setup | Monthly | Order |
|---|---|---|---|---|---|---|---|
| RTX PRO 4000 GPU1 × RTX PRO 4000, 24 GB | 14c / 20t | 64 GB DDR4 | 2 × 512 GB NVMe | HelsinkiFinland | $259 | $329/mo | Order |
| RTX PRO 6000 GPU1 × RTX PRO 6000, 96 GB | 20c / 20t | 128 GB DDR5 | 2 × 1.92 TB NVMe DC | HelsinkiFinland | $1,199 | $1,499/mo | Order |
| RTX PRO 6000 Xeon GPU1 × RTX PRO 6000, 96 GB | 24c / 48t | 256 GB DDR5 ECC Reg | NVMe DC | Falkenstein | $1,799 | $2,249/mo | Order |
| 4 × RTX PRO 6000 GPU4 × RTX PRO 6000, 384 GB | 24c / 48t | 128 GB DDR5 ECC Reg | 2 × 1.92 TB NVMe DC | HelsinkiFinland | $4,319 | $5,399/mo | Order |
$329per month
$259 setup
$1,499per month
$1,199 setup
$2,249per month
$1,799 setup
$5,399per month
$4,319 setup
All GPU servers use NVIDIA RTX PRO Blackwell cards with ECC GDDR7 memory. The RTX PRO 4000 is the affordable choice for inference and image generation. The RTX PRO 6000 carries 96 GB of VRAM on one card, and the four-GPU server offers 384 GB for training and the largest open models.
Three GPU models are hosted in Helsinki, Finland, which is also in the EU. The RTX PRO 6000 Xeon is hosted in Falkenstein, Germany. GPU servers ship with Ubuntu or Debian; we can preinstall the NVIDIA driver and CUDA toolkit.
Not sure which suits you? Read how to choose a dedicated server or ask us.
The RTX PRO 6000 Xeon server is in Falkenstein, Germany. The other GPU servers are in Helsinki, Finland. Both locations are in the EU and covered by the GDPR.
Ubuntu and Debian. We can preinstall the NVIDIA driver, CUDA and Docker with the NVIDIA container toolkit on request.
GPU servers are assembled to order with expensive cards, so the upstream setup fee equals roughly one month's rent. It is charged once.
Tell us what you plan to run and we'll suggest a server. Most questions get an answer within the hour.