Usage licenses
Access to private models we host for you.
€60/mo
We set up one dedicated server per 25 people.
- Hosted, private models — your data stays yours
- Predictable license tiers
- No data sent to third-party APIs
- Early access through the waitlist
Pricing
Pricing
On-premise AI is tailored to each company. Choose how you want to work with us — we scope the details together.
We set up individual, isolated communities — one private instance per client.
Custom installation
We design, install and deploy AI on your own infrastructure.
On request
- Hardware advisory & setup
- On-premise model deployment
- Integration & fine-tuning
- Training & ongoing support
Other services
Consulting, governance, data and operations.
On request
- AI strategy & consulting
- Data engineering & governance
- Operation, support & MLOps
- Tailored to your needs
Want a sense of the numbers? Estimate your potential savings vs the cloud. Try the calculator →
Token & cost calculator
Estimate your monthly token volume and compare cloud cost with an on-premise setup.
How much will your team use AI?
Adjust assumptions
Your estimate
- Queries / month
- —
- Tokens / month
- —
- Cloud cost / month
- —
- On-premise cost / month
- —
- Estimated monthly saving
- —
- Estimated yearly saving
- —
Rough estimate for orientation only. Adjust the prices to your real figures — on-premise gives you unlimited tokens at a fixed cost.
On-premise vs commercial cloud AI
How private, on-premise AI with open-source models compares to commercial cloud APIs.
| Privonis · on-premise | Commercial cloud APIs (GPT-4o, Claude, Gemini…) | |
|---|---|---|
| Where your data lives | Your infrastructure | Third-party servers |
| Privacy | Data never leaves | Sent to the provider |
| Cost model | Fixed, you own it | Per token, variable |
| Cost at scale | Amortizes as you use it | Grows with every query |
| Tokens | Unlimited | Metered & billed |
| Customization | Full (fine-tune, your models) | Limited to the API |
| Offline / air-gapped | Yes | No |
| EU data sovereignty | Yes | Depends on provider |
| Vendor lock-in | None — open source | High |
Qualitative comparison. Commercial APIs are great for some cases — we help you choose; we are not against them.
Open-source models vs commercial licenses
Owning open-weight models versus renting access under a proprietary license.
| Open-source models | Commercial license / API | |
|---|---|---|
| Model weights | You download & own them | No access |
| License | Apache 2.0 · MIT · open | Proprietary terms of service |
| Run anywhere | Yes, your hardware | Provider only |
| Usage limits | None | Rate & quota limits |
| Price changes | You control cost | Vendor can change anytime |
| Continuity | Runs forever, even offline | Depends on the vendor |
Our commitments to you
The promises we put in writing on every engagement.
Your data never leaves
Models and data stay inside your infrastructure — always.
No vendor lock-in
Open-source models and multi-vendor hardware. You own everything.
NDA by default
We sign a confidentiality agreement before we see anything sensitive.
Free initial assessment
We scope your use case and feasibility before you commit.
EU-based, EU data
European company, European data residency, European compliance.
SLA-backed support
Clear response times and ongoing maintenance you can rely on.
Up to a large share may be fundable
There are financing lines to hire us
Many European and national programs co-fund digitalisation and AI projects. We help you identify and apply for the ones you qualify for.
Kit DigitalNext Generation EUHorizon EuropeRegional & national aid
Grants & subsidies
European, national and regional funding for AI and digitalisation.
We guide the application
We help you map your project to the right program.
Lower upfront cost
Make on-premise AI accessible for any company size.
Availability and conditions depend on each program and your eligibility. We advise; we are not a public body.
Frequently asked questions
The essentials about on-premise AI.
What does "on-premise" mean?
The AI runs on hardware inside your own infrastructure (or a private environment you control), instead of a third-party cloud.
Does my data leave the company?
No. Your prompts, documents and customer data stay within your network. That is the whole point of on-premise.
What hardware do I need?
It depends on your use case and the model size. We advise and source it — from a single GPU workstation to a multi-node cluster.
What does "unlimited tokens" mean?
Cloud APIs charge per token. When the model runs on your own hardware you can process as many tokens as you want with no per-token billing.
Is it GDPR and EU AI Act compliant?
Yes. We build governance, documentation and controls into the project to align with GDPR and the EU AI Act.
Is there financing?
Often yes. Several European and national programs co-fund AI and digitalisation. We help you find and apply.