2 × 96 GB Blackwell GPUs
Fifth-generation Tensor Cores with FP4 and FP8 support, and 192 GB of GDDR7 memory across two cards.
GPU servers · Blackwell
GPU RTX PRO dedicated server
192 GB of Blackwell GPU memory.
$2,199/mo
Same-day delivery
Excluding taxes · no setup fee · cancel anytime
Compare all 10 servers
- Unmetered 25 Gbps
- DDoS protection
- IPMI + KVM console
- 24/7 engineers
Overview
Where GPU RTX PRO fits.
GPU RTX PRO combines two NVIDIA RTX PRO 6000 Blackwell GPUs with 96 GB each, a 48-core AMD EPYC 9455P, 576 GB of DDR5-6000 ECC, two 7.68 TB NVMe Gen5 drives and a 25 Gbps port with unmetered traffic. With 192 GB of GPU memory, it serves 70B-class language models, runs fine-tuning jobs and handles heavy 3D and simulation work. It costs $2,199/mo and is usually delivered the same day.
Ideal for
- LLM inference up to 70B-class models
- Fine-tuning with LoRA
- 3D rendering and simulation
- Multi-model serving
Partition GPUs with MIG
Each RTX PRO 6000 can be split into up to four isolated instances with Multi-Instance GPU, so several models or teams can share a card.
48-core EPYC 9455P host
96 threads and 576 GB of DDR5-6000 handle batching, tokenization and data pipelines that feed both GPUs.
Specifications
Everything in the box.
Hardware
| Processor | AMD EPYC 9455PZen 5 · 3.15 / 4.4 GHz |
|---|---|
| Cores / threads | 48 cores / 96 threads |
| GPU | 2 × NVIDIA RTX PRO 6000 Blackwell 96 GBNVIDIA data-center GPU, drivers installable from the OS |
| Memory | 576 GB DDR5-6000 ECC12 channels, one DIMM per channel |
| Storage | 2 × 7.68 TB NVMe Gen5 |
| Network port | 25 GbpsUnmetered traffic, inbound and outbound |
Service
| IP addresses | 1 IPv4 + /64 IPv6Additional IPv4 on request |
|---|---|
| DDoS protection | Always on, included |
| Out-of-band access | IPMI and HTML5 KVM console |
| Operating systems | Ubuntu, Debian, AlmaLinux, Rocky Linux, Windows Server, Proxmox VEOr any custom ISO through the KVM console |
| Delivery | Same-day deliveryAssembled to order, usually delivered the same day |
| Billing | $2,199 per monthNo setup fee, no contract, excluding taxes |
Included with GPU RTX PRO
- Unmetered traffic on the 25 Gbps port
- Always-on DDoS protection
- IPMI and HTML5 KVM console
- 1 IPv4 address and a /64 IPv6 block
- Free private VLANs between your servers
- 24/7 support by engineers
- No setup fee, no contract
Availability
Deploy GPU RTX PRO in 3 locations.
New YorkNot offered here- AshburnUS East
- DallasNot offered here
- Los AngelesNot offered here
AmsterdamNetherlands
FrankfurtGermany
LondonNot offered here
SingaporeNot offered here
TokyoNot offered here
SydneyNot offered here
Assembled to order in the available cities and usually delivered the same day. Pick a city to start your order there.
Compare
How it stacks up.
| GPU serversGPU L40S | GPU servers · this serverGPU RTX PRO | Pro seriesPro 48 | |
|---|---|---|---|
| Price | $979/mo | $2,199/mo | $999/mo |
| Processor | AMD EPYC 9355P | AMD EPYC 9455P | AMD EPYC 9455P |
| Cores / threads | 32 / 64 | 48 / 96 | 48 / 96 |
| GPU | 1 × L40S · 48 GB | 2 × RTX PRO 6000 · 96 GB | None |
| Memory | 256 GB DDR5-6000 ECC | 576 GB DDR5-6000 ECC | 576 GB DDR5-6000 ECC |
| Storage | 2 × 3.84 TB NVMe Gen5 | 2 × 7.68 TB NVMe Gen5 | 2 × 7.68 TB NVMe Gen5 |
| Port | 25 Gbps unmetered | 25 Gbps unmetered | 25 Gbps unmetered |
| Delivery | Same day | Same day | In minutes |
| Locations | 4 | 3 | 8 |
| Action | View GPU L40S | Deploy | View Pro 48 |
FAQ
GPU RTX PRO, answered.
Which models fit on two 96 GB GPUs?
As an estimate, a 70B model needs about 70 GB for weights at FP8, so it fits on one card with about 26 GB left for the KV cache, and you can run one copy per GPU. At BF16, the same model needs about 140 GB and is split across both GPUs with tensor parallelism. FP4 support on Blackwell halves weight memory again for models quantized to that format.
Are the two GPUs linked with NVLink?
No. The RTX PRO 6000 Blackwell has no NVLink, so the two cards exchange data over PCIe 5.0. That suits running one model per GPU or tensor parallelism across the two cards; jobs that depend on very high GPU-to-GPU bandwidth across many GPUs are better served by NVLink-based systems.
Which driver and CUDA version do I need?
Blackwell GPUs need a recent NVIDIA driver and CUDA 12.8 or later; earlier CUDA releases do not support the architecture. Install the driver from NVIDIA's repository on Ubuntu 24.04 LTS or another supported distribution, and use framework builds compiled for CUDA 12.8 or newer, which current PyTorch releases provide.
Where is GPU RTX PRO available and how fast is delivery?
GPU RTX PRO is available in Ashburn, Amsterdam and Frankfurt for $2,199/mo. It is assembled to order and usually delivered the same day. For other GPU counts or a fleet, ask for a custom build or volume quote.
Deploy GPU RTX PRO for $2,199/mo.
Usually delivered the same day, with unmetered traffic, DDoS protection and IPMI included.
- No setup fees
- Unmetered traffic
- Cancel anytime