Swiss Institutions Unveil Open-Source AI Model Apertus, a Transparent and Multilingual Solution for Future AI Innovation
A groundbreaking open-source AI model, christened Apertus, has been launched by a consortium of Swiss institutions – EPFL, ETH Zurich, and the Swiss National Supercomputing Centre (CSCS). Named after the Latin word for ‘open’, this model signifies transparency in its design and training process, making it accessible to all.
Apertus offers a versatile platform for developers and organizations to create chatbots, translation tools, or education-focused applications. The model can be obtained directly from Hugging Face or through Swisscom, a strategic partner of the initiative. Two versions of Apertus are available – an 8-billion-parameter model and a larger 70-billion-parameter version, both distributed under a permissive open-source license, enabling use in research, education, and commercial projects.
Unlike other AI systems that disclose limited details, Apertus offers complete transparency – its architecture, training data, and documentation are all accessible for scrutiny.
Martin Jaggi, Professor of Machine Learning at EPFL and a member of the Steering Committee of the Swiss AI Initiative, stated, “With this release, we aim to offer a roadmap for developing a trustworthy, sovereign, and inclusive AI model.” He further added that Apertus will be routinely updated by a team of engineers and researchers from CSCS, ETH Zurich, and EPFL.
Thomas Schulthess, Director of CSCS and Professor at ETH Zurich, described Apertus as “a catalyst for innovation and a means to strengthen AI expertise in research, society, and industry.” He emphasized that the project is not just a technology transfer from research to product but an endeavor to create long-term infrastructure.
The training process involved processing 15 trillion tokens across more than 1,000 languages, with approximately 40% of the data in non-English languages. Apertus supports languages often excluded from other AI models, such as Swiss German and Romansh.
Imanol Schlag, technical lead of the project and Research Scientist at ETH Zurich, noted, “Apertus is built with public welfare in mind. It stands among the few fully open AI models at this scale and is the first to embody multilingualism, transparency, and compliance as fundamental principles.”
Swisscom has already integrated Apertus into its sovereign AI platform, demonstrating its commitment to fostering a secure and responsible AI ecosystem that benefits the public interest and strengthens Switzerland’s digital sovereignty.
While downloading Apertus is straightforward for experienced users, practical application requires servers, cloud resources, or dedicated interfaces. Developers can test Apertus during the Swiss {ai} Weeks, which run until October 5, 2025. Hackathon participants will gain access via a Swisscom-hosted interface. Swisscom business customers can also begin using the model immediately through the company’s AI platform. For international users, Apertus will be accessible through the Public AI Inference Utility.
Joshua Tan, Lead Maintainer of the Public AI Inference Utility, commented, “Currently, Apertus is the leading public AI model – a model developed by public institutions, for the public interest. It serves as our best testament that AI can function as public infrastructure, like highways, water, or electricity.”
Under the open-source license, training data, model weights, and intermediate checkpoints are all available. The model’s development adhered to Swiss data protection rules, Swiss copyright law, and the transparency requirements of the EU AI Act.
The dataset was sourced from publicly-available information, with personal data removed and website opt-out requests honored. Ethical guidelines were also applied to exclude unwanted content before training commenced.
Antoine Bosselut, Professor at EPFL and Co-Lead of the Swiss AI Initiative, expressed, “Apertus demonstrates that generative AI can be both potent and open.” He further stated, “The launch of Apertus marks the beginning of a journey, a long-term commitment to open, trustworthy, and sovereign AI foundations, for the public good worldwide.”
Future updates aim to broaden the model family, enhance efficiency, and develop domain-specific tools for sectors like law, health, climate, and education – all while maintaining stringent standards of transparency.
