Switzerland is stepping into the global artificial intelligence (AI) spotlight with a major announcement: the release of its first open, multilingual large language model (LLM). Developed by leading Swiss institutions, including Eidgenössische Technische Hochschule Zürich (ETH Zurich) and École Polytechnique Fédérale de Lausanne (EPFL), the initiative represents a significant step toward AI sovereignty, transparency, and linguistic inclusivity.

Set for release by the end of summer 2025, the model embraces full open-source access and relies on one of the world’s most powerful supercomputers. With this move, Switzerland aims to become a serious contender in a space long dominated by private tech giants.

Built for Scale, Designed for Access

The team will release the model in two sizes: one with 8 billion parameters and another with 70 billion. Both versions prioritize accessibility and openness. Unlike many proprietary alternatives, this LLM draws from a massive multilingual dataset that includes over 1,000 languages. While English remains dominant in the training data, the project puts a strong emphasis on linguistic diversity, making it especially relevant for global use in translation, localization, and public-sector communication.

Engineers trained the model on the Alps supercomputer at the Swiss National Supercomputing Center. This high-performance infrastructure runs entirely on carbon-neutral energy, aligning the project with sustainability goals while ensuring technical excellence.

A Clear Stand on AI Ethics and Sovereignty

This initiative extends beyond technology — it carries a political message. By offering open access to the model’s weights, code, and documentation, Switzerland asserts its vision for AI: one rooted in openness, accountability, and legal responsibility.

For the language industry, the implications are clear. A multilingual model built on ethical data practices can reshape the foundation for AI-driven communication.