EU Institutional LLM released as an open model

The EU Institutional LLM is a large language model – the technology behind today's AI systems – developed by DG Translation to better serve the EU's 24 official languages. 

Enhanced with high-quality multilingual data from the EU institutions, it is designed to offer more inclusive language coverage and to better understand EU-specific content and terminology.

The EU Institutional LLM is now available for download so that others can use it or further build on it – access is open to any legal entity based in an EU country. This is a significant milestone in the European Commission's pioneering work on fair AI for a multilingual Europe. 

Preventing digital language extinction 

Many large language models underperform on low-resource languages – those with fewer speakers and less linguistic data available. 

As well as offering better performance in EU official languages, the aim is to make the EU Institutional LLM a better option for EU topics and more suitable for EU public administrations, small businesses, academia and non-governmental organisations.

Built on European technology 

DG Translation's expert engineers  have built on existing European open LLM using European supercomputing infrastructure and high-quality multilingual data from the EU institutions. 
Read more about how the model was built 

Using the EU institutional LLM 

The EU Institutional LLM already powers eSummary, one of our AI-based multilingual tools.

Any EU-based legal entity can download the model from the European Language Data Space:

• EU Institutional LLM v1 base – continually pretrained multilingual base model

• EU Institutional LLM v1 instruct – instruction-tuned version of the base model