Infomaniak Launches Its Open AI API: A Swiss Alternative to Cloud Giants

Dictated by Brocasm and corrected and translated by mistral-14b

Introduction

Infomaniak, the Swiss web hosting giant, has taken a significant step forward by launching its Open AI API service with new open-source models. This offering stands out for its open-source approach, competitive pricing, and a local touch in a market dominated by American players.

For developers, startups, and businesses looking for a European, transparent, and ethical alternative to solutions like OpenAI, Anthropic, or Mistral AI, this announcement deserves your attention. Here’s why.

🔍 What Does Infomaniak Offer?

Infomaniak now provides API access to open-source AI models, hosted on its Swiss infrastructure. Here are the key points to remember:

• Open-source models: Compatibility with popular models (Llama, Mistral, etc.), with the possibility of integrating others in the future.
• Local hosting: Data remains in Switzerland, a major advantage for users concerned about data sovereignty and GDPR compliance.
• Simplified integration: An API designed to be intuitive, with clear documentation and SDKs available for the most common languages (Python, JavaScript, etc.).
• First million tokens free: A generous offer to test the platform without commitment.

💰 Pricing: Competitive and Transparent

Infomaniak positions its pricing slightly above OpenRouter but below the most expensive providers for equivalent models. Here’s a quick comparison (based on information available in May 2026):

Source: Indicative prices based on information available in May 2026. Exchange rate used: 1 CHF ≈ 1.10 USD.

👉 The plus: No hidden fees, pay-as-you-go billing, and a policy of total transparency.

🧪 First Tests: A Promising but Improvable Experience

I had the opportunity to test the service by connecting Hermès, my personal AI assistant, to Infomaniak’s API. Here are my first impressions:

✅ Stability: No noticeable latency, even with simultaneous requests.
✅ Speed: Response times comparable to OpenRouter for similar models.
✅ Simplicity: Integration was smooth, thanks to well-structured documentation and ready-to-use code examples.
✅ Responsive support: Infomaniak’s team quickly answered my technical questions.

⚠️ Areas for improvement:

• Minimalist interface: The interface is very simple, and usage data is currently limited to:
  • The date of requests.
  • The number of hits (API calls).
  • The cost incurred by the API (per day and per model).
• Lack of advanced details: Unlike OpenRouter, there is no access to statistics such as the number of tokens per request, average response times, or detailed logs.
• Beta version: Since the service is still in beta, I hope these features will be added soon.

Note: I plan to conduct further tests in the coming days to evaluate scalability and long-term reliability.

🌍 Why Choose Infomaniak?

1. Data Sovereignty

• 100% Swiss hosting, compliant with GDPR and local laws.
• No data transfer to foreign servers.

2. Ethics and Transparency

• No “black box”: the models used are open-source, and their operation is documented.
• Commitment to responsible and sustainable AI.

3. Flexibility

• Ability to deploy your own models or use those offered by Infomaniak.
• API compatible with existing tools (such as Hermès, Opencode, etc.).

4. Local Support

• A technical team based in Switzerland, responsive and accessible.

🔮 Who Is This Service For?

• Developers: To easily integrate AI features into their applications.
• Startups: To benefit from a scalable and cost-effective solution.
• Businesses: For a secure alternative that complies with European regulations.
• Researchers: To access open-source models without restrictions.

📢 My Opinion

Infomaniak confirms its reputation as an innovator by offering a local, competitive, and open-source solution. If in-depth tests confirm stability and performance, this service could well become a must-have for European users.

👉 I will integrate it into my testing stack alongside OpenRouter and other providers to objectively compare performance, costs, and reliability. The lack of detailed logs is a point to watch, but the beta version suggests improvements are on the way. Stay tuned!

📊 Next Steps: Upcoming Benchmarks

To go further, I will conduct latency and throughput benchmarks this week. The results will be integrated into this article as soon as they are available. Here’s what I will measure:

• Average latency per model (in ms).
• Throughput (requests per second).
• Stability under variable loads.

🔹 Update planned: End of the week.

🔗 Useful Links

• Infomaniak Official Site - AI Services

💬 What About You?

Have you already tested Infomaniak’s API? What are your thoughts on the interface or missing features? Share your experiences in the comments or on LinkedIn / Reddit!

This article will be updated regularly with benchmark results. Stay tuned!