Apertus Swiss AI Assistant

Modified: 2025-11-12

Developed by the Swiss AI Initiative through a collaboration between ETH Zurich, EPFL, and several Swiss universities, Apertus is designed to bring AI assistance out of the private sector and into the public sphere. Trained with the power of the Alps supercomputer at CSCS, this Swiss AI model was built to process over 15 trillion tokens covering a wide range of languages and topics.

The Public AI Inference Utility recognizes that trust and safety are essential for a model to effect positive social change. By adopting privacy measures that are both robust and user-centered, Apertus and the Utility aim to set a high standard for responsible AI development, fostering a community where transparency and user control are at the heart of AI-driven services.

Key Features of Apertus

The Swiss AI model can operate in multiple languages, showcasing exceptional proficiency in Swiss national tongues such as German, French, Italian, Romansh, and Swiss German dialects. Its knowledge cutoff as of March 2024 continues to serve as a robust foundation for general knowledge, while maintaining adherence to legal, ethical, and democratic principles.

Operational Principles: Human-Centered AI

Operating under principles such as accuracy and transparency, Apertus ensures that its AI-driven responses are not only factual but also understandable and respectful. Its adaptability in addressing query complexity, from straightforward to complex questions, is a testament to its commitment to matching human needs with AI capabilities.

Swiss Values Embodied in Apertus

Grounded firmly in Swiss democratic and federalist traditions, Apertus seeks to uphold principles of consensus-building and cultural sensitivity. Its role in the Public AI Inference Utility contributes to a broader vision of accessible AI, a resource that respects human autonomy while promoting collaborative, open innovation.

Why Apertus Matters in the Public Domain

Moved by public institutions and operated through the Public AI Inference Utility, Apertus becomes a model for AI transparency and public access. It supports local and international perspectives, bridging the gap between AI research and real-world applications. As a sovereign AI model from Switzerland, it represents a significant step in challenging the dominance of commercially-driven, closed-source AI.

Links