The SOOFI (Sovereign Open Source Foundation Models) project has officially launched, marking an important step toward open, trustworthy, and European-aligned AI language models. The initiative aims to develop large-scale foundation and reasoning models that reflect European values, multilingual diversity, and regulatory requirements, while supporting a wide range of public and industrial applications.

L3S Research Center, a CAIRNE Gold Member, plays a key role in SOOFI through its contributions to multilingual datasets, safety and evaluation frameworks, and reasoning benchmarks. This work helps ensure that the resulting models are not only competitive, but also reliable, transparent, and aligned with Europe’s human-centred approach to AI.

The launch of SOOFI highlights the importance of open collaboration and shared research infrastructure in strengthening Europe’s AI sovereignty and reducing dependence on non-European foundation models.

Find out more about the initiative here.