Lingua Custodia becomes Dragon LLM

Lingua Custodia is proud to announce the launch of Dragon LLM, effective Wednesday, October 15, 2025.

Dragon LLM represents a groundbreaking advancement in artificial intelligence, marking a pivotal evolution in our mission beyond automated translation. This new model embodies our vision for a sovereign, innovative, and accountable European AI ecosystem.

Developed by our R&D team, Dragon LLM is the first modern AI architecture trained exclusively on EuroHPC’s European supercomputers. This milestone was made possible through the European Commission’s Large AI Grand Challenge, in which Dragon LLM was selected as a winner in June 2024, granting it access to Europe’s most powerful computational resources.

A New Generation of European AI

Dragon LLM introduces a hybrid architecture that redefines efficiency, accessibility, and sustainability in large language models. Unlike current dominant models (e.g., GPT, Claude, Llama), which rely on the energy-intensive Transformer architecture, Dragon is optimized for long-form reasoning and complex text processing, delivering equivalent accuracy with significantly reduced computational cost and energy consumption.

Empowering European Businesses

Dragon’s architecture integrates cutting-edge scientific research from global institutions, operationalizing key innovations from the past two years. It enables secure, scalable deployment of generative AI across enterprises, particularly SMEs, on modest private infrastructure.

This achievement underscores Europe’s growing capacity to foster innovation in AI, proving that European startups and small-to-medium enterprises can compete globally with revolutionary, homegrown AI technologies.

The full press release is available here: Press Release.

We welcome inquiries regarding our strategic shift and are eager to discuss our pipeline of generative AI innovations.