A group of researchers from Switzerland is gearing up to launch one of the world’s most open large language models (LLMs) later this summer. The project, led by ETH Zurich, EPFL, and the Swiss National Supercomputing Centre, is built to compete with big commercial AI models. But it promises transparency and multilingual support on a level not seen before.
However, one could ask what’s so special about this forthcoming Swiss model. So the reason is that everything about this LLM will be public, including the source code, training data, and model weights. The model has been trained on public Swiss infrastructure using the “Alps” supercomputer, which relies on green energy and boasts thousands of specialized AI chips.
Researchers say the model is designed to support more than 1,000 languages, including many that are often left out by major tech companies. About 40% of the model’s data comes from non-English sources. The team is prioritizing data privacy, copyright compliance, and transparency throughout the project.
Furthermore, two versions of this model are expected to arrive, including an 8-billion parameter model for everyday tasks and a larger 70-billion parameter model for complex AI applications. Both will be available under the Apache 2.0 license, making them free to use, modify, and share.
The model is finishing its last round of testing and is expected to roll out publicly before the end of summer. This means we can expect the launch within the next few weeks or by the end of September 2025. The Swiss initiative aims to set a new standard for openness and trust in AI development worldwide.
That’s a wrap on Swiss LLM for now. Want the latest Tech and AI updates before everyone else? Join us on WhatsApp, we drop news, deep dives, listicles, and more, straight to your phone.
