Skip to content
← Models
Open weightsSwiss AI Initiative/10 trustVerified

Apertus

Switzerland's fully open 8B/70B sovereign LLM — weights, code, and training data public.

Visit official page →

About

Apertus is Switzerland's first large-scale public language model, released on September 2, 2025 by the Swiss AI Initiative — a collaboration between EPFL, ETH Zurich, and the Swiss National Supercomputing Centre (CSCS). It ships in 8B and 70B dense variants under the Apache 2.0 license, and its openness goes unusually far: architecture, weights, training code, intermediate checkpoints, and the training-data pipeline are all public.

Trained on more than 15 trillion tokens spanning 1,500+ languages (40% non-English, including Romansh and Swiss German) on the Alps supercomputer's 10,000+ NVIDIA GH200 chips, it was built for GDPR and EU AI Act compliance from the start. Reception noted genuine transparency leadership alongside a candid caveat: the first iteration trailed frontier models on raw capability, and chat is a secondary use case — Apertus is infrastructure for sovereign, auditable AI rather than a ChatGPT rival.

Specs

Context window

Pricing

Free quota

API access

Weights

Open weights

Released

September 2, 2025

Benchmarks

Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.

Use cases

sovereign-aimultilingual-nlpresearchcompliant-deployment

Trust score — 8/10

/10 trustVerified

Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 80; Trending score 76/100 from researcher assessment.

More in the apertus family

Sources