Pricing · platform
Dedicated NVIDIA B200 GPUs, in Spain. Truly sovereign.
Full GPUs or MIG fractions, by the hour or reserved, plus inference on open models (GLM 5.3, Qwen3.8, Nemotron, ALIA) on NVIDIA B200 in Madrid. Your code and your data never leave the country.
180 GB
HBM3e per B200 GPU
1/8
Smallest fraction, hardware-isolated with MIG
€8.00
Per hour for a full B200, no commitment, excl. VAT
100%
Data processed and stored in Spain
How the slice works
Your slice is yours. By hardware. All the time.
We use NVIDIA Multi-Instance GPU (MIG): the B200 is physically partitioned into isolated instances. Each slice has its own compute, HBM3e memory, cache, and bandwidth. You don't compete with anyone for cycles. Your 1/4 is always your 1/4, even when the rest of the GPU is maxed out.
- Hardware isolation (not time-slicing, not virtualization): SMs, memory, and cache are physically separated between slices.
- Guaranteed bandwidth: your share of HBM3e doesn't slow down if other customers saturate their slice.
- Reserved 24/7 (6-month minimum), or by the hour with no commitment when you hit traffic peaks.
Each slice = isolated SMs + HBM3e + L2 cache + NVDEC/NVENC · no noisy neighbor
Why a dedicated slice
Your AI, inside your perimeter. No exceptions.
With a public API, your prompts train the next model and your data crosses three continents before returning. With a dedicated slice in Madrid, nothing leaves. Same model, isolated environment, compliance by design.
What happens in your slice, stays in your slice
Privacy, compliance and sovereignty built in. Not add-ons.
Data in Spain, 100%
Prompts, embeddings and responses never leave Madrid. Zero CLOUD Act exposure, zero US sub-processors, zero international transfers for Legal to sign.
Private model and context
Your B200 slice is yours with MIG hardware isolation. Your inputs don't train the next model, and your throughput doesn't depend on the tenant next door. Nobody else touches your weights.
ISO 27001 + ENS Media included
Your auditor gets the certificates directly. Your CISO closes due diligence without expanding the SoA. No extra audits, no ambiguous DPAs.
Dedicated endpoint, not shared
Private HTTPS with mTLS + VPN, only reachable from your IPs. No enforced rate limits, no inference queues. The latency is yours, 24/7.
InfiniBand co-location
Your pod, your storage and your tokens live in the same rack, wired over InfiniBand. Fewer hops, lower latency, zero cross-region egress. Your multi-step agent doesn't choke on the network.
And time pays for itself too
Operational savings are a side effect. They still cover the slice 3× over.
Team
10 devs
× 80 €/h
Idle time
30 min/day
× 220 workdays
Annual cost lost
88,000 €
1100 h/year idle
Annual 1/4 slice
26,280 €
1/4 slice reserved
Return on time
+ 61,720 €/year
3× the sliceThe real reason to switch is sovereignty and compliance. Recovered time is the bonus that wins over Finance.
Your data, your model, your latency. And your team stops waiting, too.
Mix them
Three modes. You build the combo.
Reserve a slice for your own model. Add hourly bursts when traffic spikes. And pull Token Factory tokens for a big model when you don't want to manage the GPU. All in the same cluster, all sovereign, each line billed separately, no surprises.
€/month · dedicated GPU
Fixed monthly fee for a GPU or MIG fraction dedicated 24/7, with guaranteed capacity. 6-month minimum. Ideal for dev teams and stable production.
Best for stable production
€/hour · pay as you use
Spin up a fraction or a full GPU and pay hourly until you shut it down. No commitment, no reservation; capacity subject to cluster availability.
Best for spikes and POCs
€/1M tokens · Token Factory
Pay only for the tokens the model generates. No GPU management. Call the private HTTPS endpoint from your app. Perfect for variable-scale production inference.
Best for product inference
GPU Compute with MIG
From 1/8 of a GPU to a full 8-GPU node. Always dedicated.
MIG fractions of 1/8, 1/4 and 1/2, full GPUs and an 8× B200 HGX node for training and enterprise workloads. Same platform, same contract: go from prototype to production without migrating.
01 / Mini
1/8
B200
OCR, voice, embeddings and small models (up to ~14B) · proofs of concept. The entry point.
Reserved
€1,095/month
On-demand
€1.50/hour
02 / Slice
1/4
B200
Coding assistant for 3-5 devs · light fine-tuning · models up to ~30B with large context.
Reserved
€2,190/month
On-demand
€3.00/hour
03 / Half
1/2
B200
Real production for 8-12 devs · 70B inference at native FP8 precision · training of small-to-mid models.
Reserved
€4,380/month
On-demand
€6.00/hour
04 / Full B200
1 ×
B200
72B models at FP8 full precision · high-throughput inference for teams of 15+ devs · distributed training.
Reserved
€5,840/month
On-demand
€8.00/hour
05 / HGX Cluster
8 ×B200
8× B200 with intra-node NVLink 5 and inter-node InfiniBand NDR · foundation model training · inference at scale · dedicated enterprise compliance.
Reserved
€46,720/month
€64.00/hour
Prices excl. VAT. On-demand: hourly, no commitment, subject to availability. Reserved: guaranteed exclusive capacity, 6-month minimum; the monthly price equals 730 hours.
Committing for 12 months or more? Talk to us and we will prepare a proposal. →
Token Factory
The latest open-source models. Served fast.
Open models served from our B200s in Spain, many in NVFP4, at the same prices as GPU Flow (gpuflow.ai), our self-serve platform. Your prompts and context never leave the country.
| Model | Params | Context | Input / 1M | Output / 1M |
|---|---|---|---|---|
| Z.aiGLM 5.3 NVFP4New | 753B · MoE | 1M | €1.206 | €3.789 |
| QwenQwen3.8Coding | 27B | 96k | €0.10 | €0.30 |
| QwenQwen3 | 14B | 32k | €0.20 | €1.00 |
| QwenQwen3.5Fast | 9B | 64k | €0.10 | €0.15 |
| BSCALIA Instruct | 40B | 160k | €0.20 | €2.00 |
| GoogleGemma 4 | 26B · MoE | 32k | €0.12 | €0.30 |
| NVIDIANemotron 3.5 LightningFast | 30B · MoE | 128k | €0.12 | €0.30 |
| NVIDIANemotron Nano VLVision | 12B | 128k | €0.06 | €0.24 |
| QwenQwen3-VL OCRVision | 32B | 64k | €0.10 | €0.35 |
Prices in euros per million tokens, excl. VAT, pay-as-you-go. Voice: Whisper large-v3 at €0.0015 per audio minute and speech synthesis from €5 per million characters. High volume or your own model? We deploy it on a dedicated GPU; ask us.
Sandboxes
Development environments, billed per active hour.
Container with SSH, Python, PyTorch, Jupyter and coding agents, in the same cluster as inference. It pauses itself when you stop using it: idle hours don't count.
Starter
€0.15
/active hour
4 vCPU · 8 GB RAM
Pro
€0.30
/active hour
8 vCPU · 32 GB RAM
Team
€0.55
/active hour
16 vCPU · 64 GB RAM
Persistent disk billed separately, from €5/month. Prices excl. VAT.
Where your code lives
Three places. One gives you all of them.
There's no always-right option. There's one that combines speed, privacy, and capacity, and two that force a trade-off.
Local, on your machine
Maximum physical privacy (nothing leaves the device), but bounded by RAM and bandwidth. Big models don't fit or run slowly. Your laptop is unusable during inference.
Gains privacy · loses speed and capacity
Third-party API
Fast, with powerful models, but every prompt travels to someone else's servers, with variable retention policies and jurisdiction that shifts per provider. Internal compliance will cost you hours.
Gains speed · loses privacy
Dedicated cluster in Madrid
B200 cluster speed with HBM3e, latest-gen models at native precision, VM-level isolation. Prompts and code process here. 100% Spanish data residency, ISO 27001 and ENS Media certified.
Speed · privacy · capacity
All plans include
Tailored proposal
Every use case is different. Tell us what you want to do and we'll send you a concrete proposal in under 24 hours.