Switzerland’s federal institutes of technology—ETH Zurich and EPFL in Lausanne—have announced plans to launch a large language model (LLM) of their own. Unlike most of its peers, this one will be public, open-source and led by ethics and science. It is expected to be released later in 2025.

Developed under the Swiss AI Initiative, launched in December 2023, the project pools resources from over 10 Swiss institutions. The new Swiss National AI Institute is coordinating the project, which draws on more than 70 AI researchers from ETHZ and EPFL.

Training of the system will be done on the new Alps supercomputer—equipped with over 10,000 NVIDIA Grace Hopper chips.

The Swiss model will compete on a crowded field with other LLMs such as ChatGPT, Claude, Gemini and Grok. However, it will stand out in a sector dominated by secretive American and Chinese firms, and its transparency could prove influential—particularly in Europe, where policymakers fret about digital sovereignty and algorithmic opacity.

“Fully open models enable high-trust applications and are necessary for advancing research about the risks and opportunities of AI. Transparent processes also enable regulatory compliance,” said Imanol Schlag, research scientist at the ETHZ AI Center, who is leading the effort alongside EPFL AI Center faculty members and professors Antoine Bosselut and Martin Jaggi.

The model will be fully open: source code and weights will be publicly available, and the training data will be transparent and reproducible, supporting adoption across science, government, education, and the private sector. This approach is designed to foster both innovation and accountability.

A distinctive feature of the model is its capability in over 1,000 languages. “We have emphasised making the models massively multilingual from the start,” says Antoine Bosselut.

Training of the base model was done on a large text dataset in over 1,500 languages — approximately 60% English and 40% non-English languages — as well as code and mathematical data. Given the representation of content from all languages and cultures, the resulting model maintains the highest global coverage and applicability.

However, whether a model made in Switzerland can compete with the titans of Silicon Valley or Shenzhen remains to be seen.

More on this:
EPFL article (in English)

For more stories like this on Switzerland follow us on Facebook and Twitter.

Share.
Exit mobile version