By - Telekom launches the first commercial offering featuring OpenGPT-X’s Teuken-7B
- AI model ‘Made in Germany’ strengthens digital sovereignty for businesses and public authorities
- Flexible deployment options in all 24 EU official languages with top-tier security and compliance standards
With the EU project OpenGPT-X, funded by Germany’s Federal Ministry for Economic Affairs and Climate Action (BMWK), Germany is driving the development of open and trustworthy AI language models. This initiative enhances digital sovereignty and provides German authorities and institutions access to cutting-edge European generative AI (GenAI) technology. A consortium led by Fraunhofer Institutes IAIS and IIS, alongside partners such as TU Dresden, the German Research Center for Artificial Intelligence (DFKI), and Forschungszentrum Jülich, the Teuken-7B language model was developed and trained on the Jülich supercomputer JUWELS. Since November 26, the model has been available under an open-source license.
Deutsche Telekom has successfully transitioned this European language model from research to commercialization, becoming the first provider to offer Teuken-7B for commercial use. Authorities and businesses can now access a powerful, trustworthy, and open-source alternative to mainstream AI models, marking a significant step toward Germany’s digital sovereignty.
“If not now, when does Europe need more sovereign solutions? Only through joint European efforts we can create competitive alternatives to major international providers. Teuken-7B from OpenGPT-X is a flagship project ‘Made in Germany’ and a crucial component in the digital strategies of customers who prioritize sovereignty,” said Dr. Ferri Abolhassan, CEO of T-Systems and Member of the Board of Deutsche Telekom AG.
Responsible AI with the ‘Made in Germany’ Seal of Quality
OpenGPT-X emphasizes “Made in Germany” in its development of large AI language models, addressing the needs of European businesses and research institutions alike. This focus enables future-ready, data-driven business solutions. The models stand out for their multilingual capabilities, open-source accessibility, and trustworthiness, adhering to Deutsche Telekom’s ethical AI principles.
The Teuken-7B model is designed to be multilingual from the ground up, offering energy and cost efficiency through a specially developed tokenizer. It supports all 24 EU official languages and sets new benchmarks for European AI solutions with its comprehensive open-source licensing (Apache 2.0) and optimized performance.
“Providing Teuken as an open-source model has multiple advantages. Companies can tailor the model to their needs, developing specialized applications. They can also choose to run the model locally on their infrastructure or with a trusted cloud provider of their choice. This allows sensitive data to remain within the organization,” said Dr. Nicolas Flores-Herr, Project Lead at Fraunhofer IAIS.
Wide Applications and Flexible Deployment Models
Telekom’s Teuken-7B offering is tailored to meet the specific demands of European businesses and public authorities, empowering German enterprises and contractors in the growing GenAI market.
With AI Foundation Services, Telekom subsidiary T-Systems provides flexible deployment options. Businesses can operate models in ultra-secure, certified Telekom data centers in Germany, in compliance with GDPR across Europe, or even on their own infrastructure. These solutions are particularly suitable for regulated industries and highly sensitive data, meeting the most stringent security and compliance requirements.
Additionally, the Teuken-7B model is seamlessly integrated into Telekom’s standardized Business GPT product. It supports retrieval-augmented generation (RAG) applications and combines document summarization with enterprise-grade chatbot capabilities. Through a unified API, businesses can integrate the model into existing AI assistants, agents, and proprietary chatbots. AI Foundation Services also enable the development and deployment of scalable, secure generative AI applications.
