Generative artificial intelligence is revolutionizing the way we work, but it raises confidentiality concerns. Infomaniak offers AI Tools with LLM models hosted in Switzerland, allowing you to integrate AI into your applications while protecting your data.
AI Without Compromising Your Data
Using ChatGPT or Claude via their American APIs means sending your data to the USA. For companies subject to GDPR or handling sensitive data, this is a major issue.
- OpenAI/Anthropic – Data processed in the USA, subject to the Cloud Act
- Infomaniak AI Tools – Data processed in Switzerland, privacy guaranteed
Available Models
Infomaniak offers several optimized open-source LLM models:
Hosted LLM Models
- Llama 3 – Meta model, excellent performance-to-cost ratio
- Mistral – French model, highly performant in multilingual tasks
- Mixtral – MoE architecture for high-quality responses
- CodeLlama – Specialized in code generation
Sovereign and Confidential AI
Your prompts and data never leave Switzerland. No use for training third-party models.
Use Cases
Chatbots and Assistants
Create conversational assistants for your website or application:
- Automated customer support
- Intelligent FAQ
- Sales assistant
- User onboarding
Content Generation
Automate content creation:
- Product descriptions
- Blog articles
- Marketing emails
- Document summaries
Analysis and Extraction
Automatically process your documents:
- Support ticket classification
- Entity extraction (names, dates, amounts)
- Sentiment analysis
- Automatic translation
Development Assistance
Support your developers:
- Code generation
- Automatic documentation
- AI-assisted code review
- Generated unit tests
The LLM API in Practice
The API is compatible with the OpenAI format, making integration easy:
Simple Integration
Use the same libraries as for OpenAI (openai-python, langchain, etc.) by simply changing the endpoint and API key.
API Features
- Chat completions – Multi-turn conversations
- Completions – Simple text generation
- Embeddings – Vectorization for RAG
- Streaming – Real-time responses
RAG: Retrieval Augmented Generation
Combine LLMs with your own data using RAG:
- Index your documents in a vector database
- Search for relevant passages
- Augment the LLM context
- Get answers based on your data
Perfect for creating expert chatbots on your documentation, products, or internal processes.
Pricing
Billing is based on tokens (text units) consumed:
| Model | Input (1M tokens) | Output (1M tokens) |
|---|---|---|
| Llama 3 8B | ~0.20€ | ~0.30€ |
| Llama 3 70B | ~1.00€ | ~1.50€ |
| Mistral 7B | ~0.20€ | ~0.30€ |
| Mixtral 8x7B | ~0.70€ | ~1.00€ |
Advantages Over Alternatives
| Criterion | Infomaniak AI | OpenAI | Self-Hosting |
|---|---|---|---|
| Data Location | Switzerland | USA | Your Servers |
| GDPR | 100% Compliant | Risks | Compliant |
| Maintenance | Zero | Zero | Significant |
| Hardware Cost | None | None | Expensive GPUs |
| Scalability | Automatic | Automatic | Complex |
Conclusion
Infomaniak\'s AI Tools allow you to integrate generative AI into your applications while respecting the confidentiality of your data. It is the ideal solution for European companies that want to innovate without compromising their GDPR compliance.
Ready to migrate to Infomaniak?
Contact us for a free 15-minute audit. We will analyze your situation and provide you with a personalized quote.
Request a free audit