LISBON, Could 28, 2025 | Multilingual open-source initiatives EuroLLM and OpenEuroLLM have joined forces to safe 3 million GPU hours on Leonardo – considered one of Europe’s strongest supercomputers – to develop a groundbreaking artificial dataset overlaying 40 European languages.
The initiative was chosen underneath the EuroHPC AI Manufacturing unit Giant Scale name recognizing its potential to advance Europe’s management in multilingual synthetic intelligence.
On the coronary heart of this initiative is a mission to construct strategic autonomy for Europe in AI growth. By producing high-quality, ethically sourced artificial knowledge, it addresses a long-standing hole in linguistic illustration, particularly for low-resource and minority languages.
André Martins, Chief Scientific Officer at Unbabel and EuroLLM challenge co-lead stated:
“By becoming a member of forces by means of EuroLLM and OpenEuroLLM, we’re bringing collectively the analysis power and open-source ethos wanted to sort out considered one of Europe’s largest AI challenges: linguistic inclusion at scale. This challenge is about making certain Europe owns its language knowledge, displays its cultural variety, and units its personal requirements in accountable AI growth.”
The GPU allocation will energy the MultiSynt strategy, a key part of the challenge which seeks to deal with some of the persistent bottlenecks in multilingual LLM growth: the shortage of high-quality pre-training knowledge.
“This is a vital step in securing massive sufficient computing energy to construct the OpenEuroLLM’s household of open LLMs. I’m additionally glad that this has been carried out in collaboration with the skilled group from the EuroLLM challenge. The aim of this subproject is to discover multilingual artificial knowledge creation and consider their use with the intention to attain the next widespread aim: constructing high-quality multilingual LLMs for all European languages and past.” – notes Jan Hajic, Charles College, coordinator of the OpenEuroLLM challenge.
Whereas most artificial knowledge era for giant language fashions to this point has targeted on English, MultiSynt will create the primary complete multilingual artificial dataset designed particularly for pre-training. By leveraging generative fashions to reinforce and diversify present content material, it would help the broader goals of EuroLLM and OpenEuroLLM: constructing open-source, culturally grounded, and linguistically various AI for Europe.
This technique will help linguistic variety, open entry, and knowledge high quality and aligns with the broader goals of the European Fee’s Digital Decade and the AI Act.
The awarded 3 million hours mirror a powerful endorsement of the challenge’s technical benefit and strategic worth.
The initiative will probably be executed by means of phased releases of the artificial dataset.
****ENDS****
About EuroLLM
The EuroLLM challenge contains Unbabel, Instituto Superior Técnico, the College of Edinburgh, Instituto de Telecomunicações, Université Paris-Saclay, Aveni, Sorbonne College, Naver Labs, and the College of Amsterdam. Collectively they created EuroLLM-9B, a multilingual AI mannequin supporting all 24 official EU languages. Developed with help from Horizon Europe, the European Analysis Council, and EuroHPC, this open-source LLM goals to reinforce Europe’s digital sovereignty and foster AI innovation.
About OpenEuroLLM
Bringing collectively 20 of Europe’s main AI corporations, analysis establishments and EuroHPC centres, the OpenEuroLLM challenge is creating a brand new era of open supply massive language fashions for European languages. Co-funded by the European Union’s Digital Europe Programme, the challenge is laying the foundations for AI infrastructure that can improve competitiveness, resilience, and digital sovereignty.
About EuroHPC
The European Excessive Efficiency Computing Joint Enterprise (EuroHPC JU) is a joint initiative between the EU, European international locations, and personal companions to develop a world-class supercomputing ecosystem in Europe.
Media Contacts:
For extra data or interview requests, please don’t hesitate to achieve out to our media contacts under:
• Unbabel: farah.pasha.ext@unbabel.com
