OpenEuroLLM is a new initiative aimed at democratizing access to high-quality AI technologies, enhancing the competitiveness of European companies and public organizations.

Officially launched on February 1, 2025, the project is funded by the European Commission's Digital Europe Programme.

The primary goal of OpenEuroLLM is to develop powerful open-source language models that support all official languages of the European Union.

With a total budget of €52 million, the project also benefits from additional funding through the EU's STEP program, aimed at bolstering strategic technology investments.

A significant challenge for the initiative is Europe's limited computing capacity compared to the U.S. and China, although it will utilize facilities within the EuroHPC network for model training.

This initiative aligns with Europe's goals of improving digital sovereignty and competitiveness in the global AI market, especially against the backdrop of advancements by American and Chinese companies.

Coordinated by Jan Hajič from Charles University and co-led by Peter Sarlin of AMD Silo AI, the project involves 20 European research institutions and companies.

The project has received the Strategic Technologies for Europe Platform (STEP) Seal, underscoring its significance and alignment with EU standards.

While no specific timeline for the release of the language models has been disclosed, there is considerable anticipation regarding their development and potential applications.

Alongside the AI models, the project will provide comprehensive documentation, training codes, and evaluation metrics to support fine-tuning for specific industry needs.

OpenEuroLLM represents Europe's strategic response to maintain competitiveness in the AI landscape, particularly against established technologies like ChatGPT and DeepSeek.

Developing multilingual models that serve all EU member states equally presents a significant challenge, particularly for low-resource languages.