EU project releases 7B model that speaks 24 European languages

The 7-billion-parameter language model Teuken-7B is now available on Hugging Face, offering support for all 24 official European Union languages. The model comes from the EU's OpenGPT-X research project and is available as open-source. Unlike most AI language models that focus mainly on English, Teuken-7B was built from scratch with about half of its training data coming from non-English European languages. The developers say the model performs reliably across all languages it was trained on. The project team also created the European LLM Leaderboard that measures how well LLM's work across European languages, moving beyond the English-only testing that was standard before.