OpenEuroLLM

"Sovereign AI: Must We Own the Model to Own Our AI Future?"

Tomorrow, 28th Oct. at #EMNLP26, represented by Andrey Kutuzov, we will be sharing the OpenEuroLLM perspective on sovereign AI at this very interesting panel within the @WiNLPWorkshop.

Join the discussion at 11:00!

Today at @emnlpmeeting, @maxidahl & @TiedemannJoerg & Andrey Kutuzov presented MultiSynt, a multilingual synthetic dataset to support low-resource EU languages on behalf of OpenEuroLLM and EuroLLM supported by the AI Factories initiative. 🫶🇪🇺

Just in case you missed our poster:

Hello Budapest! We are at #EMNLP2026!

If you're around, we'd be more than happy to chat about the interesting nuances of building open, multilingual foundation models.

Say 👋 to the #OpenEuroLLM team!🇪🇺

📸 Anna, Amaru, Zihao, Jörg, Tomazs, Andrey, Jan, Max, Otto and Niclas

A series of technical reports from our project, like the one just posted by our colleague @kleiaaro, will try to provide answers to the many questions that LLM training teams cope with around the globe.

This one is about training recipes 👇

Our 9B experimental WIP model now named

🎼 PRELUDE

is still training but we are sharing intermediate checkpoints for transparency:

🛑 WIP research artifact.
🛑 NOT cooled down yet.
✔️Include validation across 36 languages.

Full WIP model card👇huggingface.co/openeurollm/pr…