Switzerland Unveils Apertus: A Fully Open AI Model for Global Use
A revolutionary open-source AI model, Apertus, designed by Swiss institutions, is poised to reshape research and application development.
Amsterdam, California, & London, [Date] – A group of leading Swiss research institutions, EPFL, ETH Zurich, and CSCS (Swiss National Supercomputing Centre), have released Apertus, a groundbreaking open AI model built for transparency and global accessibility. Apertus, Latin for “open,” embodies its core principle by making every aspect of its design and training process, from architecture to data, publicly available.
Open Access, Enhanced Capabilities. Developers and organizations can now leverage Apertus to build innovative applications like chatbots, translation tools, and educational platforms. Downloadable directly from Hugging Face or accessible via strategic partner Swisscom, Apertus comes in two powerful versions: an 8 billion parameter model and a more robust 70 billion parameter model. Both are released under a permissive open-source license, encouraging widespread adoption across research, education, and commercial ventures.
A Paradigm Shift in AI Development. Unlike many proprietary AI systems, Apertus stands out with its comprehensive open approach. Its architecture, training data, and detailed documentation are available for scrutiny, fostering trust and collaboration within the AI community. “Apertus provides a blueprint for developing trustworthy, sovereign, and inclusive AI models,” says Martin Jaggi, Professor of Machine Learning at EPFL and member of the Swiss AI Initiative Steering Committee. The model will continuously be updated by a dedicated team of engineers and researchers from CSCS, ETH Zurich, and EPFL.
Multilingual Mastery and Global Impact. Apertus’s training was meticulously crafted using 15 trillion tokens and data spanning over 1,000 languages, with a notable emphasis on non-English languages—including Swiss German and Romansh. This commitment to multilingualism positions Apertus at the forefront of accessible AI models. “Apertus is built for the public good, embodying multilingualism, transparency, and compliance as foundational design principles,” states Imanol Schlag, technical lead of the project and Research Scientist at ETH Zurich. Swisscom’s deployment of Apertus on its sovereign AI platform further emphasizes the project’s commitment to shaping a secure and responsible AI ecosystem.
Real-World Access and Testing The Apertus model, downloadable via Hugging Face, will enhance practical use cases—requiring servers, cloud resources, or dedicated interfaces. Experienced developers and hackathon participants can test Apertus during the Swiss AI Weeks (through October 5, 2025), supported by Swisscom. Swisscom business customers can utilize the model through their AI platform today, with international users accessing it through the Public AI Inference Utility. Apertus is solidifying its position as a leading public model for the public good, mirroring essential public infrastructure.
Transparency, Ethics, and the Future of Open AI. Apertus adheres strictly to Swiss data protection rules, Swiss copyright law, EU AI Act requirements, and ethical guidelines—ensuring comprehensive transparency during the training process. The training data was restricted to publicly available information, with personal data meticulously filtered out. This commitment to transparency and ethical conduct ensures the model’s rigorous development, laying the foundation for future iterations while upholding these critical principles.
“Apertus demonstrates that generative AI can be powerful and open,” adds Antoine Bosselut, Professor at EPFL and Co-Lead of the Swiss AI Initiative. “Its release marks a new chapter in fostering long-term commitments to open, trustworthy, sovereign AI foundations for global benefit.” Future updates will focus on model expansion, improved efficiency, and the development of specialized tools for various domains, maintaining the highest standards of transparency throughout.
About Apertus: Apertus is a fully open AI model developed by EPFL, ETH Zurich, and CSCS (Swiss National Supercomputing Centre). It is available through Hugging Face and via Swisscom’s platform, emphasizing transparency, multilingual support, and responsible AI development.
Learn More: Visit the Apertus model page on Hugging Face https://www.artificialintelligence-news.com/news/switzerland-releases-its-own-fully-open-ai-model/.
Keywords: Apertus, open AI model, Switzerland, AI research, open-source, Hugging Face, Swiss AI Initiative, multilingual, AI development, Generative AI, AI ethics, responsible AI, sovereign AI
(Image: [Placeholder for Apertus logo or relevant image])
(Note: Please replace bracketed information like the date and image placeholder)