Apertus
Switzerland's first large-scale sovereign LLM: fully open weights, data, and recipes, trained on 15T tokens across 1,000+ languages.
Overview
Apertus is Switzerland's first large-scale open sovereign language model, built by EPFL, ETH Zurich, and the Swiss National Supercomputing Centre (CSCS) under the Swiss AI Initiative. Released in September 2025 in 8B and 70B sizes, it is fully open: weights, training data, and training recipes are all public. It was trained on 15 trillion tokens spanning more than 1,000 languages, with about 40% non-English data, on the Alps supercomputer in Lugano.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 24 Jul 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
ASEAN Perspective
Apertus in Southeast Asia
ASEAN-region availability and pricing notes coming soon. Drop the editorial team a note via /contact/ if you can supply local context (Singapore/Malaysia/Indonesia/Thailand/Vietnam).
What this is for: A fully transparent, multilingual open LLM (8B and 70B) built as European sovereign infrastructure, with weights, data, and training recipes all published.
Who this is for: Researchers, public institutions, and developers who need an auditable, openly licensed multilingual model, including for underrepresented languages.
Availability: Free and fully open; downloadable on Hugging Face and hosted via Swisscom and the Public AI Inference Utility. Built by EPFL, ETH Zurich, and CSCS in Switzerland.
What people say
Apertus is received first as a governance and transparency milestone rather than a leaderboard contender. Coverage from ETH Zurich, swissinfo, and Wikipedia consistently praises it as Switzerland's first large-scale fully open model, releasing weights, training data, and recipes for 8B and 70B variants trained on 15 trillion tokens across 1,000+ languages with roughly 40% non-English data. For researchers, public institutions, and anyone needing an auditable, EU/Swiss-compliance-friendly multilingual model, that openness and its respect for crawler opt-outs are the whole point, and its multilingual coverage of underrepresented languages is a real differentiator.
The candid caveat, acknowledged even by supporters, is performance. Independent benchmarking and commentary note that Apertus does not beat proprietary models across the board, and its ethically curated data, while widely applauded, is flagged as potentially reducing diversity and hurting niche or specialized tasks, though the team argues degradation is minimal for general use. Observers also point to the project's relative lack of large-scale training experience and its dependence on centralized public infrastructure and community support. It is explicitly built for transparency and sovereignty, not to top every leaderboard, and readers should judge it on that basis rather than as a frontier chatbot.
Summary of public user & expert reviews, compiled by RECATOOLS.
About this listing
This entry was compiled from publicly available data including Apertus's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Apertus unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Apertus directly →
Spotted something out of date? Suggest an update →
Alternatives to Apertus
More in LLMs & Chat
