Europe’s most powerful open-source multilingual language model Apertus debuted at the Zurich Hackathon. Built by developers at the ETH Zurich on the ALPS supercomputer, it was trained in 1,000 languages under strict EU data and AI regulations—an industry first. Unlike U.S. models, the sources used to train Apertus are also fully transparent, with copyright issues addressed. The long-term aim of the project is to boost Europe’s digital sovereignty.