Switzerland has made history by releasing Apertus, the world’s first fully transparent AI model developed by public institutions. Created collaboratively by EPFL, ETH Zurich, and the Swiss National Supercomputing Centre, this groundbreaking initiative offers complete openness in architecture, training data, and documentation.
Unlike proprietary AI systems, Apertus provides unprecedented transparency with its 8-billion and 70-billion parameter versions available under permissive open-source licenses. The model supports over 1,000 languages, including underrepresented ones like Swiss German and Romansh, with 40% of training data in non-English languages.
Trained on 15 trillion tokens while adhering to Swiss data protection laws and EU AI Act requirements, Apertus serves as public infrastructure for research, education, and commercial applications. Swisscom has already integrated it into their sovereign AI platform, demonstrating Switzerland’s commitment to digital sovereignty and responsible AI development for global public benefit.