I haven't tried it for anything myself yet. The paper provides several benchmarks. The emphasis during training was on multi-language support (over 1800 languages are represented in its pre-training data, which is 40% non-English) and non-copyrighted training data... and the benchmarks seem to suffer for it.

https://arxiv.org/abs/2509.14233

Yes it’s not bad, although it’s not meant to be a chatbot, post training is limited, so it won’t feel as smooth as TOTL of course. The number of supported languages is mind boggling.

Focus was on open data, languages and auditability.

Their loss function is fancy, not sure about the effects

Not really. It's because the target audience is more academic/scientific rather than the Swiss population at large. In the latter case, it would be in the local languages. The law is relatively clear for this. English is not accepted in Switzerland as a replacement language for the "local" ones, although many people can speak or at least understand some English.

Staff nationality of Swiss higher education institutions:

- Universities: 55% Swiss, 45% foreign - Universities of applied sciences: 75% Swiss, 25% foreign - Universities of teacher education: 87% Swiss, 13% foreign - Professors: 49% Swiss, 51% foreign - PhDs/scientific collaborators: 30% Swiss, 70% foreign - Professors of ETH Zurich: 31% Swiss, 69% foreign

What if I told you there’s this thing in 2026 called an LLM that can translate between any two languages with high fidelity for free, and you just clicked a single button in your browser to use it

Not quite: it's a collab between both ETHZ (Zürich, German speaking) and EPFL (Lausanne, French speaking). According to the website, the actual hardware is distributed all over the country (including in the Italian part).

And the other wars ;)