OdiModel
An assistant with the best open models and voice in Spain's official languages, on our platform in Spain.
Sovereign NVIDIA Blackwell compute · Madrid
AI on your data, without it leaving Spain.
Dedicated NVIDIA B200 GPUs, inference on open models and consulting for companies and public bodies that cannot send their data to an API outside the EU. Our own hardware, our own team, a contract under Spanish law.
No commitment on hourly billing · Public prices in euros, excl. VAT
Three ways to work with us
One platform, three ways in.
Start with the one that fits where you are and move to another without changing provider or contract.
A · InfrastructureCompanies
Dedicated NVIDIA B200 GPUs
Whole or in hardware-isolated fractions. Hourly to validate, reserved for production.
- From 1/8 of a GPU to 8 × B200 nodes
- SSH access and custom images
- Exascaler parallel storage
from€1.50/h
See configurations →B · InferenceSMEs and developers
Open models by API
GLM, Qwen, Nemotron, Gemma and ALIA served from Spain. Compatible with the OpenAI and Anthropic SDKs.
- Pay per million tokens
- SSH sandboxes billed per active hour
- Sign up in minutes on GPU Flow
from€0.06/M
Go to GPU Flow ↗C · ConsultingLeadership and CTOs
Buy, rent or hybrid
We size your workload on our cluster before you buy anything. From DGX Spark to HGX B200, and if you buy, we operate it.
- TCO of all three paths with real numbers
- GPU datacenter design
- Managed operations
first session45 min, free
Book the session →Where data sovereignty matters most
For sectors that cannot send their data abroad.
Law firms
Long case files, scanned documents and attorney-client privilege.
Industry and intellectual property
Designs, patents, code and manufacturing processes that are your competitive edge, processed without leaving Spain.
Public administration
A platform certified ENS Medium on a datacenter certified ENS High.
Healthcare
Patient records and data processed in an isolated environment in Spain.
Banking and insurance
Workloads under DORA and NIS2, with traceability and a single audited provider.
Integrators and partners
Our GPUs with your network and your cybersecurity, white-label too.
Running today
Not a slide deck. You can see it working.
iQuantum · Granada
Smart-city solutions programme: AI applied to public services.
Granada City Council
GPU Solutions-UGR Chair
Chair in Accelerated Computing Infrastructure and Sustainable AI with the University of Granada.
How we work
First you see it and measure it. Then you decide.
- 01
We walk you through it
45-minute session with an engineer: use case, models, volume and budget.
Week 0 - 02
We show you
Visit to the datacenter in Madrid and a live demo on B200.
Week 1 - 03
You try it
An hourly GPU fraction and a sandbox. First agents running within days.
Weeks 1 and 2 - 04
You scale
Review at 30 days with real usage. Reserve, add GPUs or move to a full node.
Month 1 - 05
We operate it
If you would rather buy, we size the NVIDIA hardware and run it for you.
When it makes sense
Public prices
You know what you pay before talking to anyone.
GPU Solutions list prices in euros, excl. VAT. Reservations have a 6-month minimum; hourly billing has no commitment.
| Configuration | Memory | €/hour | €/month |
|---|---|---|---|
| 1/8 of a B200 | ≈ 23 GB | €1.50 | €1,095 |
| 1/4 of a B200 | 45 GB | €3.00 | €2,190 |
| 1/2 of a B200 | 90 GB | €6.00 | €4,380 |
| Full B200 | 180 GB | €8.00 | €5,840 |
| 8 × B200 node | 8 × 180 GB | €64.00 | €46,720 |
Frequently asked questions
The real questions CTOs, CISOs and platform leads ask us.
- 01
How much does an NVIDIA B200 GPU cost at GPU Solutions?
€8.00 per hour for a full B200 (180 GB), excl. VAT and with no commitment. Reserved, it costs €5,840 per month with a 6-month minimum. Fractions start at €1.50 per hour (1/8 of a GPU).
- 02
Where is my data processed and stored?
In a Tier III datacenter in Madrid, on hardware owned by GPU Solutions. Your data never leaves Spain and there are no subprocessors outside the European Union.
- 03
What security certifications do you hold?
The platform is certified UNE-EN ISO/IEC 27001:2023 and Spain's National Security Framework (ENS) at Medium level, by EQA. The datacenter holds ENS High, ISO/IEC 27001, ISO 9001 and ISO 50001.
- 04
What is a MIG GPU fraction?
A hardware-isolated part of the GPU with its own memory and compute, with no neighbours slowing it down. A B200 supports up to 7 instances of 23 GB, 4 of 45 GB or 2 of 90 GB.
- 05
Can I use open models without managing GPUs?
Yes, by API on GPU Flow, our self-serve platform, from €0.06 per million input tokens. It is compatible with the OpenAI and Anthropic SDKs: you only change the endpoint and the key.
- 06
How long until I am in production?
72 hours for a standard dedicated environment. Custom setups, such as a full 8-GPU bare-metal node, take one to two weeks.
- 07
Can you serve the Spanish public sector?
Yes. The platform is certified ENS Medium on a datacenter certified ENS High, and we are a Spanish operator, which simplifies public procurement.
- 08
What happens if I want to leave?
You export your models, weights and data in open formats. Hourly billing has no commitment and no penalty; reserved capacity has a 6-month minimum.
Insights
Blog & Research
What we think, what we research, what we know.
Infrastructure6 Oct 2026
How much does an NVIDIA B200 cost in Spain? Buy, rent or hybrid
What renting a B200 in Spain costs compared with the big cloud providers, and everything behind owning one: 24/7 operations, failures, datacenter, compliance and obsolescence.
Read article →
Regulation6 Oct 2026
ENS and AI: Medium vs High category for AI workloads in Spain
For anyone buying or auditing AI infrastructure in Spain's public sector or a regulated industry: how a system's ENS category is set, what the High category adds, what each certificate covers and what to ask for in a tender.
Sovereignty13 Jul 2026
Ontology and sovereignty: who defines what your data means
The ontology is the layer that gives your data meaning, and the one a third party would rather manage for you. It closes the AI sovereignty trilogy: after weights and learning, meaning. And why in Spain and Europe that layer is also a question of jurisdiction.
Sovereignty12 Jul 2026
You pay for AI twice: the Reverse Information Paradox
Satya Nadella just named AI's second invoice: you pay it with your company's knowledge. We explain what the Reverse Information Paradox is, why it connects with what Karp already said, and what a normal company can do to protect itself without giving up the best models.
Executive briefings
Take it to your board
Two executive guides, direct download. The conversation your board needs to have, prepared by you, no lawyers in the room.
NIS2 + AI · May 2026
NIS2 compliance without lawyers in the room
Regulatory framework, the seven questions for your AI vendors, the governance the regulator looks at first, and a three-horizon action plan. Reviewed by external legal counsel.
Download guide →
PDF · 10 pages · 14 min
Sovereign stack · Technical
The combo your engineers actually need
Pod, B200 slice, Exascaler storage, tokens. What each piece is, how they fit together, three typical combos with ballpark pricing, and why it matters they live on the same rack over InfiniBand.
Download guide →
PDF · 14 pages · 10 min
Contact
Let's talk
This isn't a generic form. A real person reads it, and responds in under 24 hours.
Or reach out directly
contact@gpusolutions.aicontact@gpusolutions.ai
▸ Basic data protection information
Controller: BIAI Technology Project S.L. (CIF B75473223)
Purpose: respond to your enquiry and, where applicable, manage your commercial request.
Legal basis: your explicit consent when submitting this form.
Recipients: no data is transferred to third parties unless legally required. Resend (EU transactional email provider) processes the sending.
Rights: access, rectification, erasure, objection, portability and restriction by writing to contact@gpusolutions.ai