Sovereign AI on demand
Access the best open source AI models via OpenAI-compatible APIs, while keeping your sensitive data in Europe. No queries are stored, and the energy that powers our AI infrastructure is 100% renewable and recovered for heating purposes.
Limitless possibilities with total control over your data
LLM assistant
Create multimodal assistants capable of understanding queries, automatically breaking down tasks, interacting to gather information and improving productivity. They can translate languages, summarise content, analyse feelings, answer questions and much more.
Semantic search
Compare, group and index your data with embedding templates. Ideal for content recommendation.
Function calling to connect your models to your applications
Use the MCP to transform a query into an action: consult a database in real time, execute code to automate a process or call an API to enrich your applications.
Image generation
Illustrate your apps, dashboards or content by creating unique visuals from simple prompts.
Voice recognition
Automate note-taking, index media or help the hearing-impaired by transcribing audio files into text.
Access the best open-source models via an OpenAI-compatible API
Get your API key
Select the right model for your needs
Get 1M free credits, then pay as you go
Why Infomaniak for your AI projects?
Total sovereignty
Our AI and GPU services are powered from the heart of Europe in our data centers located in Switzerland, with no dependence on foreign players.
Technological independence
Our AI services are based on the best open source models, guaranteeing you transparency and technological independence.
Privacy
What you entrust to the AI belongs to you. Your queries are neither recorded nor used to train models or improve our services.
Compliant with local standards
With Infomaniak, you simplify your data governance with the guarantee of full compliance with the FADP and GDPR.
The world's greenest AI infrastructure
We invest heavily in R&D to reduce the environmental impact of artificial intelligence.
100% local renewable energy
In addition to purchasing exclusively local renewable electricity, Infomaniak builds its own solar power plants using European panels with a low carbon footprint. The aim is to cover 50% of our needs through self-generated energy by 2030.
100% energy recovered to heat households
Play
The renewable energy that powers our AI services is fully recovered to heat 6’000 homes in winter and provide the equivalent of 20’000 showers per day in summer. The GPUs are cooled without dedicated air-conditioning and use only a small quantity of water in the event of a heatwave, thanks to the adiabatic principle.
Certified B Corp
With an average PUE of less than 1.1, a WUE of 0.015 and carbon offsetting totalling 200% of our emissions, we are building a sovereign and sustainable European cloud designed for future generations. This commitment is based on strict, internationally recognised standards.
Discover the world of AI at Infomaniak
GPU instances for your customised AI needs
Select your GPUs, adapt the type of deployment and optimise your training or inferences with VLLM. Renowned for their energy efficiency, our infrastructures offer you all the flexibility and performance you need for your AI projects.
Euria, our ethical AI for the general public
Benefit from the power of AI directly integrated into your productivity tools to communicate, store, share and collaborate.
Do you have any questions or doubts?
We’ve compiled the questions we’re often asked – see below.
Is it possible to train AI Tools with customised data?
We don’t currently offer the option of training our AI Tools with customised data.Can I use AI Tools to sell services with my company?
Yes, AI Tools can be incorporated into your own applications, which you are free to monetise.Can I generate or interpret images using AI Tools?
The possibility of generating or interpreting images with AI Tools is currently being analysed. We will offer you this functionality when it is fully operational and also guarantees total control of your data.Can I transcribe an audio file into text using AI Tools?
Yes, AI Tools is a multi-model product including Whisper, that lets you transcribe audio files into text.How many words can AI Tools interpret per prompt?
Our proposed LLM templates have different limitations due to variations in the complexity and size of the templates. However, you can find the maximum prompts size accepted by our templates here- What is a token?Tokens are units that divide up the text before the API analyses it, either as characters or as a set of words. In general, a token corresponds to 4 characters in English or 3/4 of a word, depending on the language.
What's the difference between an incoming and an outgoing token?
Incoming tokens are requests made to the AI, and outgoing tokens are responses generated by the AI.Is Infomaniak's AI connected to the Internet?
No, the models we propose are based on a gigantic dataset, which does not need to be continuously fed with new information.Which AI model does AI Tools use?
- We use several Open Source templates to offer a wide range of tools with Ai Tools, such as Mixtral for text generation and Whisper for audio transcription.
- You can find the full list of models we offer here
How does AI Tools billing work?
You can use the AI immediately and track your consumption in real time. You will be billed at the end of the month according to the number of tokens consumed. When you finalise the creation of your account, we'll ask you to confirm your identity by registering a valid payment card, thereby reinforcing our anti-fraud measures.Does AI Tools use OpenAI standards?
Yes, the various models we offer are based on Open AI standards to facilitate the transition to and use of our AI Tools.- How does per-minute billing work?Transcription or image generation models are billed per minute of generation. This time depends on several parameters, such as the size of the file to be transcribed (for transcription) or the quality of the image and the number of images generated simultaneously (for images).
It is this generation time that is billed. For example, an image generated in 4 seconds using SDXL-Lightning will be billed in proportion to the indicated price, i.e. CHF 0.02.