Apertus – Open Foundation Model for Sovereign AI

	Apertus – Open Foundation Model for Sovereign AI(apertvs.ai)
	419 points by T-A 14 hours ago \| 140 comments
	tl;dr: Apertus is a fully open foundation model from the Swiss AI Initiative (EPFL, ETH Zurich, CSCS), releasing weights, training data, code, and methods under reproducible terms. Available in 8B and 70B parameter sizes, it's trained on 1000+ languages and designed for EU AI Act compliance, including opt-out respect, PII removal, and memorization prevention. Swisscom is a strategic partner.
	HN Discussion: ↓Other fully open LLMs already exist and may outperform Apertus ↓Project moves slowly like a committee and won't produce competitive models ↓Model quality is poor based on testing (strawberry test, language hallucinations, copyright claims) ↑Real value is the team's experience gained for future iterations ↑Open foundation models threaten commercial AI labs and represent the future