What OpenGPT-X does

OpenGPT-X is a publicly funded German research consortium developing multilingual large language models "Made in Germany" for European organizational use. Operated under the leadership of Fraunhofer IAIS (Sankt Augustin) and Fraunhofer IIS, the project brings together ten partners spanning research (DFKI, Jülich Supercomputing Centre, TU Dresden ZIH), infrastructure and cloud (IONOS, Aleph Alpha), and end-user sectors (WDR public broadcasting, ControlExpert insurance/automotive), with the consortium coordinated under the Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. legal entity.

The core technical deliverable is Teuken 7B, a 7-billion-parameter multilingual LLM trained from the ground up on 4 trillion tokens, with over 50% non-English data drawn from 23 European countries. A custom multilingual tokenizer optimized for all 24 EU official languages reduces tokenization overhead for European languages — German text incurs only 22% more compute than English, versus materially higher surcharges on English-centric tokenizers. Training was conducted on the JUWELS Booster supercomputer using up to 512 NVIDIA A100 GPUs, and the project publishes the European LLM Leaderboard for multilingual benchmark transparency. Supporting assets include a three-part LLM Workbook and a 35-application Use Case Library for enterprise adoption.

The project's business model is grant-funded rather than commercial: the German Federal Ministry for Economic Affairs and Climate Action (BMWK) provided approximately €14 million under the Gaia-X funding competition for the January 2022 to March 2025 period. The Teuken 7B model is released as open source under CC BY-NC 4.0 via Hugging Face, with no disclosed commercial pricing. Commercial deployment is enabled indirectly through partners — IONOS distributes Teuken 7B via its AI Model Hub and Deutsche Telekom became the first company to offer commercial services built on the model — but no licensing revenue accrues to OpenGPT-X itself. The project is positioned around European digital sovereignty and serves European enterprises, public broadcasters, research institutions, and public administration organizations seeking multilingual, EU-aligned AI alternatives to US and Chinese providers.

Name

OpenGPT-X

Legal name

Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V.

Website

https://opengpt-x.de

Company type

Private

Founded year

2022

Operating status

Operating

Headcount range

11–50 employees

Short description

OpenGPT-X is a German publicly funded research consortium developing Teuken 7B, a multilingual open-source LLM trained across all 24 EU languages, serving European enterprises, public broadcasters, and research institutions seeking sovereign AI infrastructure.

Ownership category

akta.pro rank

Product category

Foundation Language Models

NAICS

Software Publishers (5132), Custom Computer Programming Services (541511)

SIC

Services-Prepackaged Software (7372), Services-Computer Programming Services (7371)

akta.pro primary industry

Foundation Model Developers (LLM/Multimodal Model Labs) (HDAAACAA)

akta.pro secondary industries

Open-Source Model Ecosystems & Model Marketplaces (HDAAACAM), Enterprise Foundation Model Integration & APIs (Connectors, Governance, Deployment) (HDAAACAO)

Keywords

• Multilingual large language models
• Open source AI
• European language AI
• Foundation models
• Natural language processing

Headquarters

HQ city

Sankt Augustin

HQ country

Germany

HQ region

Europe

Offices2 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

Technology or R&D, Infrastructure, Personnel, Marketing or Sales, Operations

Revenue model

1. Research Funding: OpenGPT-X is a publicly funded research project. The German Federal Ministry for Economic Affairs and Climate Action (BMWK) funded the project from January 2022 to March 2025 with approximately 14 million euros as part of the Gaia-X funding competition. No commercial revenue model exists as the models are released as open source.

Pricing tiers

Go-to-market motion2 records

Distribution channels4 records

Marketing channels5 records

Core offering

OpenGPT-X develops and trains large multilingual AI language models "Made in Germany", released as open source. The core offering is the Teuken 7B family of 7-billion parameter foundation models supporting all 24 EU languages via a custom multilingual tokenizer, trained on 4 trillion tokens with over 50% non-English data. The project also provides the European LLM Leaderboard for multilingual benchmarking, the LLM Workbook for educational purposes, and an LLM Use Case Library cataloging 35 business applications.

Product overview

OpenGPT-X is a German research project developing large AI language models "Made in Germany" characterized by versatility, trustworthiness, multilinguality, and openness. The core product is the Teuken 7B family of multilingual LLMs (7 billion parameters, 24 EU languages), supplemented by the European LLM Leaderboard for multilingual benchmarking, an LLM Workbook for educational resources, and an LLM Use Case Library cataloging 35 business applications. All models are released as open source under CC BY-NC 4.0 license.

Differentiator

Problem solved

Functional benefit

Brands

• Teuken 7B: Multilingual open source AI language model for Europe, instruction-tuned and trained in all 24 EU languages. Released in November 2024, it is a 7 billion parameter model designed for European needs.

Products and services

• Teuken 7B A multilingual large language model with 7 billion parameters, trained on 4 trillion tokens with a custom multilingual tokenizer optimized for all 24 EU languages. Available as base model and instruction-tuned variants (v0.4 and v0.6 releases) for European organizations requiring AI capabilities across multiple European languages.
• LLM Workbook A three-part reference guide providing resources and examples to understand the key features of large AI language models, covering technology, applications, and limitations of LLMs for businesses and researchers adopting generative AI.
• European LLM Leaderboard A ranking system that evaluates multilingual language models across European languages, enabling comparison of model performance in nearly all official EU languages with up to 70 billion parameters using translated benchmarks (ARC, HellaSwag, MMLU, GSM8K, TruthfulQA, Belebele, FLORES-200).
• LLM Use Case Library A catalog of 35 realizable business applications for large language models across various industries including financial services, healthcare, manufacturing, and public administration, showcasing practical enterprise deployment scenarios.

Named customers3 records

Segments3 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

AI capability5 records

Feature4 records

Partnerships

Ten partnerships are on record, tiered core and secondary.

• Fraunhofer IAIS (Fraunhofer-Institut für Intelligente Analyse- und Informationssysteme)coreStrategic or Co-development PartnerConsortium lead together with Fraunhofer IIS. Responsible for organizational leadership, training of functional AI language models, and development of a GAIA-X node. Dr. Nicolas Flores-Herr serves as OpenGPT-X project lead. Located in Sankt Augustin near Bonn with 300+ employees.
• Fraunhofer IIS (Fraunhofer-Institut für Integrierte Schaltungen)coreStrategic or Co-development PartnerConsortium co-lead alongside Fraunhofer IAIS. Contributes expertise in audio and media technologies, speech signal processing, and speech assistance systems. Based in Erlangen with over 30 years of experience in audio standards including mp3 and AAC.
• Aleph Alpha GmbHcoreTechnology or IntegrationOne of few companies worldwide with extensive experience training large language models. Contributes code and expertise in data selection, model parallelization, and robust evaluation of training progress. Contact: Dr. Jan Therhaag, Research Manager.
• IONOS SEcoreTechnology or IntegrationProvides highly scalable cloud/GPU computing capacity. Creates foundation for operating an MLOps platform and supports deployment of a GAIA-X compliant OpenGPT-X node. Also contributes to interoperability efforts. Contact: Rainer Sträter and Maria Barros Weiss, Head of Digital Ecosystems.
• Jülich Supercomputing Centre (JSC), Forschungszentrum Jülich GmbHcoreTechnology or IntegrationContributes high-performance computing expertise. Operates the JUWELS system, Europe's most powerful supercomputer. Focus areas: methods for scaling language models on supercomputers, optimization on hundreds of GPU compute nodes, and identifying requirements for future HPC architectures.
• Zentrum für Informationsdienste und Hochleistungsrechnen (ZIH), TU DresdencoreTechnology or IntegrationProvides HPC resources and investigates performance aspects of language models. Examines parallel efficiency and energy consumption during model training, with high savings potential for large language models.
• Deutsches Forschungszentrum für Künstliche Intelligenz (DFKI)coreStrategic or Co-development PartnerGermany's leading application-oriented AI research institution with 1,300+ employees. Contributes research on GPT3-like language models and speech services, GAIA-X node provision and interoperability with other AI platforms, and analysis of sustainable usage and business models.
• ControlExpert GmbHsecondaryImplementation/ SI/ Consulting PartnerTechnology-driven claims management provider with 800+ employees worldwide processing 14M+ documents annually. Develops two use cases: digital assistant for online damage reporting and AI-powered verification of claims documents for automotive insurance sector.
• Westdeutscher Rundfunk (WDR)secondaryStrategic or Co-development PartnerPublic broadcasting company for North Rhine-Westphalia serving 18 million people. Testing language models in daily operations for mediathek/audiothek, journalistic work support, and multimedia production processes.
• Akademie für Künstliche Intelligenz (AKI) gGmbH im KI BundesverbandsecondaryGTM or Marketing PartnerGermany's largest AI network representing 400+ innovative SMEs, startups, and entrepreneurs. Responsible for networking the project with AI community and SMEs nationally and Europe-wide, project communication, and dissemination of results to stakeholders.

Scale indicators6 records

Recent moves7 records

Expansion highlights5 records

Others

• Hugging Face: Platform and ecosystem for hosting, fine-tuning, and distributing open-source AI models. OpenGPT-X distributes Teuken 7B and hosts the European LLM Leaderboard on Hugging Face, making it a critical distribution and ecosystem partner rather than a direct competitor.

Emerging players

• AI21 Labs: Foundation model developer behind Jurassic-2 and Jamba models, offering enterprise-focused LLMs with multilingual capabilities. Comparable as a smaller, specialized foundation-model lab competing for enterprise AI budgets outside the US hyperscaler orbit.
• Silo AI: European AI lab (acquired by AMD) developing foundation models including Poro and Viking with a focus on European languages and sovereign AI. Comparable to OpenGPT-X in its multilingual European-language focus and European-origin positioning.
• Stability AI: Open-source generative AI model developer (Stable LM, Stable Diffusion) with open-weight releases and a community-driven distribution model. Comparable as an open-source foundation model developer with similar licensing and ecosystem dynamics, though focused more on multimodal outputs.
• Cohere: Enterprise-focused foundation model provider (Command, Embed, Rerank) with multilingual coverage including European languages. Comparable as a non-hyperscaler LLM developer targeting regulated enterprise and public-sector customers.
• Technology Innovation Institute (Falcon): UAE-based developer of the Falcon open-weight LLM family, comparable as a non-US sovereign open-source LLM initiative targeting multilingual and enterprise use cases in regulated markets.

Direct peers

• Mistral AI: European (French) foundation model developer offering open-weight and commercial LLMs (Mistral 7B, Mixtral, Mistral Large) with multilingual support. Direct competitor to OpenGPT-X in the European open-source LLM category with significantly more capital and a larger commercial organization.
• Aleph Alpha: German foundation model developer (Luminous, Pharia) focused on European sovereign AI for enterprise and government. Direct overlap with OpenGPT-X on European LLM positioning, language coverage, and target customers; Aleph Alpha is also an OpenGPT-X consortium partner.
• BigScience BLOOM: Open multilingual large language model (176B parameters) trained by a large international research consortium to support 46 natural languages. Shares the European/public-funded research-consortium model and multilingual-first approach with OpenGPT-X.

Broad incumbents

• Meta Llama (Meta AI): Meta's open-weight LLM family (Llama 2/3/3.1) is the dominant open-source foundation model benchmark OpenGPT-X compares against. Teuken 7B is explicitly positioned to compete with Llama-3.1-8B on multilingual efficiency, making Meta the primary reference incumbent.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat4 records

Key risks6 records

Key highlights7 records

Customer concentration

• LinkedInlinkedin.com/company/opengpt-x
• Xx.com/OpenGPTX

Revenue estimate

Valuation estimate

Number of profiles

Profiles1 record

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about OpenGPT-X