Switzerland has a unique opportunity to lead in Artifical Intelligence, thanks in part to Apertus: a competitive large language model (LLM) developed by the Swiss AI Initiative that is fully open and reproducible, proving that advanced AI can be developed in an ethical and sovereign way.
We are now starting the Swiss Public Inference Utility (SPIU) as a cooperative, to support and sustain the mission of Apertus in Switzerland for years to come from the side of civic tech. We are open to individuals and businesses that are learning, building, investing, and evolving institutions with public AI.
Our mission is to (1) create new opportunities in public AI in Switzerland; and (2) accelerate the adoption of Apertus and related projects. We aim to work in partnership with the Swiss AI Initiative, with universities, governments, and research centers. Our main focus areas are explored here.
Universal Access to Knowledge
An LLM is a neural network technology trained in an ethical and compliant way on vast amounts of multilingual data within a scientific initiative. It is designed to offer insights, answers, and creative assistance on a wide range of topics. Apertus is a powerful LLM component in a platform, on top of which a multitude of research and commercial entities can grow.
Large language technologies, when developed as public service tools, can revolutionize access to information. Public AI is the notion of a civic tech or government policy program that helps to ensure fair universal access to AI, and "assert the public’s role as a key stakeholder in AI technology" [Vanderbilt 2025]. As a civic tech movement, Public AI aims to reduce reliance on proprietary models. It prioritizes sovereignty and neutrality to ensure unbiased services where users understand data handling and governance transparently. To this end, we run benchmarks, develop literature, consult civil society organizations and governments, and recommend open and sovereign tools.
Building more Inclusive AI
To build social impact, we need to access the complete lifecycle of AI development: starting with diverse datasets (e.g., FineWeb2, Common Crawl, Wikipedia, ..) and open source hubs (e.g., OpenRouter, Hugging Face, ..). Explaining the training and remixing of LLMs, development of user interfaces and agents with a comprehensive foundation that supports diversity and accessibility, we aim to democratize AI by addressing gaps for underserved communities. Fundamental to this is the offering free and inexpensive local support for Apertus users and service providers.
An Inference Utility is such an accessible platform, helping to democratize AI for public benefit through an easy to use online interface, as well as developer kit for technical users. Since the launch on September 2, 2025, Apertus has been available in this form: thanks to donated compute credits, corporate investments, and volunteer efforts. We intend to continue supporting and raising funds for this and other endpoints.
Join the Movement
Implementing a large language model as a public service tool can lead to significant societal impacts. It can become a trusted source of information, provide clear explanations, and ensure users can interact with AI safely and informedly. This approach seeks to shift the development landscape towards more inclusivity, responsibility, and community engagement. Here are a few good reasons to join the SPIU cooperative:
- Lead the AI transformation in Switzerland: help us build an AI stack that aligns with local laws, values, culture, and jobs.
- Participate in creating a reliable, transparent AI tools that serve Swiss needs, reflect diverse regions and languages.
- Grow your business with Swiss-based, subsidized public AI infrastructure.
- Be part of a community shaping the future of AI in Switzerland.
Our focus today:
- Starting a sustainable and transparent cooperative
- Using community governance to keep momentum
- Deliver on locally relevant social impact goals
Together, let’s build Public AI!
🟢 Try Apertus on the Inference Utility: publicai.co
🔵 Fill out the Apertus User Satisfaction Survey
🤗 Join the technical community on Hugging Face
🇨🇠Help us get started: join the cooperative today!
🔖 Further Resources: