The European research project OpenGPT-X has released a large language model for artificial intelligence applications. The model, called “Teuken-7B,” is available for download on the Hugging Face platform.
OpenGPT-X is a European research and development project that started in early 2022. The project’s goal is to develop a large AI language model that meets European values, privacy standards, and linguistic diversity. Teuken-7B was trained from scratch with the 24 official languages of the EU and consists of seven billion parameters.
Until now, almost all relevant AI language models in the Western world have come from the USA. These include GPT-4 from OpenAI, Claude from the AI startup Anthropic, Grok from Elon Musk’s xAI, Llama from Meta, and Gemini from Google. Experts estimate that OpenAI’s GPT-4 has around 200 billion parameters.
The European model Teuken-7B is now available worldwide as a free, publicly funded alternative for science and business. Researchers and businesses can use the open-source model in commercial projects and integrate the code into their AI applications.
The OpenGPT-X project is led by the Fraunhofer Institutes for Intelligent Analysis and Information Systems (IAIS) and Integrated Circuits (IIS). The project also involves the TU Dresden, the Jülich Research Center, and companies like Aleph Alpha and IONOS SE.
“Our model has demonstrated its capabilities across a wide range of languages, and we hope that many will adapt or further develop the model for their work and applications,” said Stefan Wrobel, director at Fraunhofer IAIS.
The aim is to contribute to the growing demand for transparent and customizable solutions in generative artificial intelligence, both within the scientific community and in collaboration with companies from various industries.