The models that run on a Pryvan server.
Every model here runs entirely on a single rental server inside the EU — smaller variants on Core, the largest on Max. Each ships with a plain-English model card so you know exactly what you're running.
Mistral
Europe's own frontier models
Llama
The most widely used open models
Qwen
Multilingual, long context, strong at code
DeepSeek
Efficient reasoning and code
Your own model
Bring or fine-tune a private one
The catalogue expands over time. Need a specific open model on the network? Tell us on the waitlist.
Know what you're running.
No black boxes. Each model card is written in plain English, so a non-technical owner can make an informed, defensible choice.
Training data
What the model was trained on, in plain English.
Limitations & biases
Known weaknesses, so you deploy with eyes open.
Use cases
Recommended and prohibited applications.
Version history
What changed, and when.
Bring your own model, or train one on your own data.
Deploy a custom model, or use your own data to train a private one that's yours alone, isolated on vanafter compute, inside the EU, never shared.
Bring AI into your business, without giving up your data.
Join the waitlist. We're onboarding GDPR-sensitive SMEs across Europe.