The DeepSeek affair with all its uncertainties and limitations has had, as we have seen, the disruptive effect of questioning the alleged superiority of the US big tech in the field of generative artificial intelligence, and it also seems to have functioned as a wake-up call for Europe, which can and must react. It must do so both for technological independence and for geopolitical dignity at a time when the US is behaving as never before in history and the consequences of the new administration’s decisions are totally unpredictable, and when China is proving to have much more power, and not only in AI, than in the past, just look at the upheavals in the automobile industry and the latest results in terms of advanced research such as that on nuclear fusion for example.
An alarm bell that does not seem to be ringing in vain as a new initiative called OpenEuroLLM has been announced, bringing together Europe’s leading companies and research institutes in the field of AI and pooling their strengths and expertise to develop next-generation open-source language models in an unprecedented collaboration to advance European AI capabilities.
A consortium of 20 research institutes, companies and EuroHPC centres in Europe, co-ordinated by Jan Hajič (Charles University, Czechia) and co-led by Peter Sarlin (AMD Silo AI, Finland), will build a family of high-performance, multilingual, large-scale language base models for commercial, industrial and public services.
Transparent and compliant open-source models will democratise access to high-quality AI technologies and strengthen the ability of European companies to compete on the global market and public organisations to produce high-impact public services.
The OpenEuroLLM project is in line with the imperative to improve Europe’s competitiveness and digital sovereignty. The project is an excellent example of the kind of technological infrastructure needed to lower the thresholds for the development and refinement of European artificial intelligence products, demonstrating the power of transparency, openness and community involvement, values widely recognised in the European technology ecosystem.
The models will be developed within the strong European framework, ensuring alignment with European values and maintaining technological excellence. In collaboration with the open-source and open science communities such as LAION, open-sci and OpenML, and other industry experts gathered in the project’s Open Strategic Partnership Board, OpenEuroLLM will ensure that models, software, data and evaluation are completely open and can be refined and fine-tuned for the specific needs of industry and the public sector . Questi modelli multilingue performanti preservano la diversità linguistica e culturale, consentendo alle aziende europee di sviluppare prodotti e servizi di alta qualità nell’era dell’IA.
The project, which has been awarded the STEP (Strategic Technologies for Europe Platform) seal of approval, draws on the support of previous European projects and the experience of the partners and their results, including high-quality big data repositories and previously developed pilot LLMs. The consortium started its work on 1 February 2025, with funding from the European Commission under the Digital Europe Programme.
The project partners are: among the universities and research organisations the Czech Charles University which is the coordinator, the French ALT-EDIC, Eindhoven University of Technology, the German ELLIS Institute Tübingen, Fraunhofer IAIS, Research Center Juelich, the Swedish Lindholmen Science Park, the University of Helsinki, University of Oslo, University of Turku and University if Tübingen. Among the companies : GenAI AMD Silo acting as co-coordinator and based in Finland, Germany’s Aleph Alpha Research and Ellamind, France’s LightOn and Spain’s Prompsit Language Engineering. Finally, the four EuroHPC centres involved are the Barcelona Supercomputing Centre, Cineca in Italy, CSC – IT in Finland and SURF in the Netherlands. (photo by Nejc Soklič on Unsplash)
ALL RIGHTS RESERVED ©