Sovereign AI from Europe: First insights into a new model
06/23/2026The University of Würzburg is involved in the Soofi project, which is introducing a new AI model called ‘Soofi S’. It offers European companies and public authorities an independent alternative to non-European providers.
The Soofi consortium has presented initial performance results for "Soofi S", the
first building block of a European AI model family.Funded by the Federal Ministry for Economic Affairs and Energy, the project aims to build high-performance AI models on European infrastructure in order to provide independent European alternatives to AI technologies from the US and China.
The aim is to create a transparent and secure alternative to overseas offerings for companies, public authorities, start-ups and researchers. This first phase of the project introduces a model that organisations can run flexibly on their own systems.
Typical areas of application include the control of industrial processes, the analysis of large volumes of documents, and assistance with writing programming code. Because the system has been trained on a large volume of text, it operates very quickly whilst consuming little energy. The focus is particularly on the German and English languages.
The Julius-Maximilians-Universität Würzburg (JMU) is involved in the project. A team led by Professor Andreas Hotho, Head of the Chair of Computer Science X (Data Science) and of CAIDAS, contributed to the initial pre-training and the subsequent specialisation of the models during the first phase of the project. “A particular focus of our work was on the quality and performance of the models in the German language,” said Hotho. Also involved is the team of JMU Professor Fotis Janidis, who is working on the specific data required for training and evaluating the AI’s logical inferences.
Independence for the European economy
"Whoever controls the foundation models controls a central part of future digital value creation and strengthens their sovereignty and resilience. With Soofi, we are building an open foundation on which businesses, SMEs and the public sector can develop transparent AI applications based on their own data, without becoming permanently dependent on individual non-European models," says Jörg Bienert, Managing Director of the Center for Sovereign AI, German AI Association.
A key aspect of the project is the disclosure of exactly how it works. The partners involved not only provide the finished model, but also publish documentation on the training methods and data flows used. This makes the technology verifiable and adaptable for public authorities and businesses.
Tailored to industrial processes
The technical implementation takes place on a cloud in Munich. Comparative performance tests have already shown that the European model can easily keep pace with international systems of a similar scale.
"Soofi S is not intended as yet another general-purpose chatbot, but as a technical foundation for industrial AI. What matters is that Soofi S performs not only well in benchmarks, but can be deployed reliably, efficiently and transparently in production," says Nicolas Flores-Herr, Technical Project Lead for Soofi and Team Lead at Fraunhofer IAIS.
Strong partners from academia and research
Behind the consortium lies a network of German research institutions, universities and AI companies. The initiative is coordinated by the German AI Association. The participating institutions also include the University of Würzburg with its CAIDAS research centre, represented by Professor Andreas Hotho, Professor Fotis Janidis, Jan Pfister and Julia Wunderle. Also involved are institutes of the Fraunhofer Society, the German Research Centre for Artificial Intelligence, as well as other universities and specialist firms.
The system is currently being tested in practice in collaboration with industry in order to gather valuable experience for its further development directly from the day-to-day realities of business. Interested companies and organisations can get in touch to arrange a trial.
Contact
Prof. Dr Andreas Hotho, Head of the Chair of Computer Science X (Data Science), T +49 931 31-88453, andreas.hotho@uni-wuerzburg.de
Lara Lawniczak, German AI Association, T +49 1520 5370752, lara.lawniczak@ki-verband.de
