# Run AI models on European infrastructure

> Tide runs the AI models behind your projects on European infrastructure: open models, processing inside the EU, and a tailored model of your own on request.

Canonical: https://www.waterglass.ai/en/products/tide/
Language: English (Deutsch: https://www.waterglass.ai/de/produkte/tide.md)
Company: Waterglass FlexCo, Börseplatz 1/3/6, 1010 Vienna, Austria · hi@waterglass.ai

## Waterglass Tide: Your AI model on European infrastructure

We run open AI models in European datacentres and, on request, develop a model of your own from them, tailored to your use case.

## For business customers: Open AI models for your business

### Processing inside the EU

The models run in European datacentres. Compute, object storage and data residency stay inside the EU, and Waterglass is an Austrian company based in Vienna.

### Operation compliant with the EU AI Act and GDPR

We work with open model families such as Mistral, DeepSeek, GLM and Kimi on our own infrastructure. That gives you ideal preconditions for the GDPR and the EU AI Act.

### The right model for the task

A smaller model can be trained from a large one so that it masters your tailored use case. That is possible on less compute and means a lower running cost.

### One point of contact

You need no team of your own to run open models or to have them tailored to you. With Waterglass you get all of it from a single source.

## Waterglass Tide: Questions & answers

**Where are the models run?**

In European datacentres, with a focus on sustainable energy sources. Waterglass, the operator, is an Austrian company based in Vienna and is therefore itself subject to European law.

**Which models can we use?**

Open model families such as OpenAI, Mistral, DeepSeek, GLM and Kimi. Our [Waves](https://www.waterglass.ai/en/products/waves/) platform additionally offers models from Anthropic, OpenAI and Google, some of them with processing inside the EU. Which combination makes sense depends on your use case.

**What is a distilled model?**

A large model acts as the "teacher" for a smaller "student" model. The student learns exactly one specific task and gets by on far less compute than a large, complex model. We offer this process on request and settle the terms with you beforehand.

**Do we have to have you build our AI project?**

Usually yes. We take on delivery and operation together, because the two can hardly be separated. If you already run an AI application, though, we can move its operation onto open models on Waterglass Tide.

**What does hosting my open AI model cost?**

Ongoing operation is billed by token usage. On request we put together an offer for your specific use case, including the volume you expect.

## Let's talk about your use case

Tell us what you have in mind and which data is involved. We will tell you whether and how it can be run in Europe.

### How can we help?

Book a no-obligation intro call and we will answer your open questions about open models on European infrastructure.