Switzerland launches Apertus, world’s first open-source national AI model

Switzerland has launched Apertus, an open-source national Large Language Model developed by public institutions including EPFL, ETH Zurich, and the Swiss National Supercomputing Centre. The move positions Switzerland as the first country to offer a fully transparent AI model as public infrastructure, potentially setting a new standard for how nations approach AI development and data sovereignty.

What you should know: Apertus represents a completely open approach to AI development, with full transparency across its entire training process.

• Users can inspect any part of the model’s training, access comprehensive documentation, source code, and the datasets used to build it.
• The model was designed to comply with Swiss data protection and copyright laws, making it potentially attractive for companies navigating European regulations.
• Anyone can use Apertus—researchers, hobbyists, and companies—to build chatbots, translators, educational tools, and other applications.

Key technical details: The model was trained on 15 trillion tokens across more than 1,000 languages, with significant multilingual capabilities.

• Forty percent of the training data consists of languages other than English, including Swiss German and Romansh.
• Apertus is currently available in two sizes: 8 billion and 70 billion parameters.
• The model can be accessed via Swisscom or Hugging Face.

In plain English: Think of tokens as the basic building blocks of language that AI models use to understand text—similar to how we might break down sentences into individual words and punctuation marks. Parameters are like the adjustable settings that help the AI make decisions about what words should come next, with more parameters generally meaning more sophisticated responses.

Why this matters: Switzerland’s approach directly addresses growing concerns about AI transparency and data practices in the industry.

• The model was trained only on publicly available data, with crawlers respecting machine-readable opt-out requests—a stark contrast to some AI companies that have been accused of bypassing website protocols.
• Several AI companies have faced lawsuits from news organizations and creatives for using content without permission to train their models.

What they’re saying: Proponents see Apertus as proof that AI can function as essential public infrastructure.

• “Currently, Apertus is the leading public AI model: a model built by public institutions, for the public interest,” said Joshua Tan, a leading advocate for AI as public infrastructure.
• “It is our best proof yet that AI can be a form of public infrastructure like highways, water, or electricity.”

The Swiss advantage: Local financial institutions had previously expressed strong interest in a homegrown AI solution.

• The Swiss Bankers Association said a domestic LLM would have “great long-term potential” due to its ability to comply with Switzerland’s strict data protection and bank secrecy rules.
• While Swiss banks currently use other AI models, it remains unclear whether they will transition to Apertus.