GPU servers
GPU dedicated servers with NVIDIA 24 to 96 GB.
A whole server with a dedicated NVIDIA GPU: RTX PRO 4000 Blackwell, L40S or RTX PRO 6000 Blackwell, for inference, fine-tuning, rendering and transcoding. Flat monthly price, no hourly meter.
At a glance
from $169/mo
- Configurations
- 3
- Delivery
- 1 in 24 h · 2 in 72 h
- Locations
- 3 of 10 cities
- Every server
- 10 Gbps unmetered, DDoS, IPMI
AMSFRANYC
Deploy RTX PRO 4000 Blackwell · $169/mo- 10 Gbps unmetered port
- Always-on DDoS protection
- IPMI / KVM access
- IPv4 + routed /64 IPv6
- $0 setup, cancel anytime
Configurations
3 GPU servers, one monthly price.
Filter by city to see what is available near your users. Every price includes the 10 Gbps unmetered port; the struck price is the market median for the same hardware.
| Server | CPU | Memory | Storage | Network | Locations | Price / month | Order |
|---|---|---|---|---|---|---|---|
| RTX PRO 4000 Blackwell | NVIDIA RTX PRO 4000 · 24 GBAMD Ryzen 9 9900X · 12C/24T | 64 GBDDR5 | 2×1.92 TB NVMeNVMe PCIe 4.0 | 10 GbpsUnmetered | Available in Amsterdam, Frankfurt, New York. | Deploy | |
| NVIDIA L40S | NVIDIA L40S · 48 GBAMD EPYC 9355P · 32C/64T | 256 GBDDR5 ECC | 2×3.84 TB NVMeNVMe PCIe 4.0 | 10 GbpsUnmetered | Available in Amsterdam, Frankfurt. | Deploy | |
| RTX PRO 6000 | NVIDIA RTX PRO 6000 · 96 GBAMD EPYC 9355P · 32C/64T | 384 GBDDR5 ECC | 2×3.84 TB Gen5NVMe PCIe 5.0 | 10 GbpsUnmetered | Available in Amsterdam, Frankfurt. | Deploy |
No configuration matches this location yet.
Need more IP addresses? Extra IPv4 is added at checkout, and every server includes a routed /64 IPv6. IP addresses
“Market avg.” is the median public list price of the same CPU, memory and storage at major European and US providers with their standard 1 Gbps port, setup fees and VAT excluded, checked September 2026. The 10 Gbps unmetered port is included on top.
Why GPU servers
What GPU servers are good at.
The whole GPU is yoursNo time-slicing and no vGPU: full PCIe access, your own driver version and CUDA stack.
Flat monthly priceFrom $0.23 per hour equivalent when run around the clock, with no metering and no egress bill.
Move models and datasets freelyPull hundreds of gigabytes of weights at 10 Gbps and serve inference traffic without counting gigabytes.
Balanced hostsRyzen 9 9900X or EPYC 9355P hosts with 64 to 384 GB of memory and NVMe, so the GPU is never starved.
Specifications
GPU server specifications, side by side.
Public NVIDIA and AMD specifications. Each GPU sits in its own server with the host CPU, memory and NVMe listed.
| GPU | Architecture | GPU memory | Tensor formats | Bus | Host | Storage | Price |
|---|---|---|---|---|---|---|---|
| NVIDIA RTX PRO 4000 Blackwell SFF · 24 GB | Blackwell | 24 GB GDDR7 ECC | FP4 · FP8 · BF16 | PCIe 5.0 | AMD Ryzen 9 9900X 12C/24T · 64 GB DDR5 | 2 × 1.92 TB NVMe | $169/mo $0.23/h equivalent |
| NVIDIA L40S · 48 GB | Ada Lovelace | 48 GB GDDR6 ECC | FP8 · BF16 | PCIe 4.0 | AMD EPYC 9355P 32C/64T · 256 GB DDR5 ECC | 2 × 3.84 TB NVMe | $549/mo $0.75/h equivalent |
| NVIDIA RTX PRO 6000 Blackwell · 96 GB | Blackwell | 96 GB GDDR7 ECC | FP4 · FP8 · BF16 | PCIe 5.0 | AMD EPYC 9355P 32C/64T · 384 GB DDR5 ECC | 2 × 3.84 TB NVMe Gen5 | $899/mo $1.23/h equivalent |
Use cases
Where GPU servers shine.
- AI & GPU computeDedicated NVIDIA GPUs for inference, fine-tuning and rendering.Our pickRTX PRO 6000$899/mo
- Streaming & CDNPush hundreds of terabytes a month without paying for one.Our pickRTX PRO 4000 Blackwell$169/mo
- Web & SaaSPredictable performance for apps, databases and CI runners.Our pickNVIDIA L40S$549/mo
FAQ
GPU servers: questions, answered.
Still have a question? Customers with an active server get answers from engineers in a support ticket.
Support ticketsWhich GPU do I need for a given model?
For inference, weights take about 2 bytes per parameter in FP16/BF16, 1 byte at 8-bit and about 0.5 byte at 4-bit, plus 10 to 30% for the KV cache. A 24 GB card runs 8B models in FP16 or about 32B at 4-bit; 48 GB fits about 70B at 4-bit; 96 GB fits 70B at 8-bit. Full sizing table
Are the drivers installed?
You get a clean OS with root access. Install NVIDIA’s driver and CUDA from the official repositories, or use Docker with the NVIDIA Container Toolkit. That way you choose the exact versions your stack needs.
Can I use the GPU for video transcoding?
Yes. These GPUs include NVENC and NVDEC hardware encoders, well suited to FFmpeg-based transcoding pipelines next to a 10 Gbps unmetered port.
Is GPU cryptocurrency mining allowed?
No. GPU servers are for compute, rendering and AI workloads. CPU mining on other servers is allowed under our Acceptable Use Policy.
Guides
Related guides
- RTX PRO 6000 rental in 2026: hourly cloud prices vs a monthly serverWhat an NVIDIA RTX PRO 6000 costs by the hour at the main GPU clouds and marketplaces in October 2026, the break-even point between hourly billing and a monthly server, the costs the hourly rate leaves out, and when hourly still wins.
- gpt-oss-120b and gpt-oss-20b hardware requirements: VRAM and GPUsWhat OpenAI’s gpt-oss-120b and gpt-oss-20b need to run: weight sizes from the model card, the memory each token of context takes, single-GPU serving with vLLM and Ollama, documented GPU support, and which 24, 48 and 96 GB cards fit which model.
- The 2026 DRAM squeeze: what it means for dedicated server buyersServer DRAM contract prices rose at record quarterly rates from late 2025 into 2026. The data, why it happened, what it does to dedicated-server pricing and how to rent only the memory you need.
Deploy RTX PRO 4000 Blackwell for $169 a month.
Or any configuration on this page, with 10 Gbps unmetered, no setup fee and no contract.
Ready in 24 hCancel anytimeBTC, ETH, XMR, USDT or SOL