TrillionLabs Awarded MSIT AI-Specialized Foundation Model Development Project

AI startup Trillion Labs has been selected as a core development institution for the Ministry of Science and ICT's ambitious 'AI-Specialized Foundation Model Project,' opening an innovative new chapter in the field of medical AI. This large-scale consortium, led by Lunit, is moving towards the bold goal of building the world's first medical science-specialized foundation model. In this project, which has the vision of 'full-cycle medical science innovation from molecules to populations,' Trillion Labs is responsible for the design and development of the core engine and learning structure of the medical science foundation model. Based on its experience in developing large language models with 70 billion parameters and its proprietary XLDA (Cross-lingual Document Attention) technology, Trillion Labs plans to implement a high-performance model that efficiently learns vast amounts of medical data and performs stable inference. In particular, Trillion Labs plays a pivotal role in the development of 'Co-Scientist (Bio-Medical Co-Scientist, BMCS),' an intelligent research partner system that allows researchers, clinicians, and pharmaceutical companies to generate and verify hypotheses in real-time with AI. Trillion Labs plans to automate the research process, from target discovery and patient cohort analysis to clinical hypothesis design and documentation, by developing the Co-Scientist's core engine. Furthermore, it will enhance the reliability of medical science research with 'Chain of Evidence' technology, which connects guidelines, clinical trials, labels, and real-world data to accurately trace and cite evidence. Through collaboration with domestic NPU company Rebellions, Trillion Labs will build South Korea's first on-premise medical AI infrastructure, ensuring medical data sovereignty while providing the best high-performance learning and inference environment. This Lunit-led consortium includes top-tier domestic AI and bio companies such as Trillion Labs, Aizen Sciences, SK Biopharm, and Kakao Healthcare, as well as multiple medical institutions including KAIST, Seoul National University research teams, and National Health Insurance Service Ilsan Hospital. The research and development will proceed in two stages, with a total project cost of 18.2 billion KRW and GPU infrastructure support worth 17.4 billion KRW. Shin Jae-min, CEO of Trillion Labs, emphasized, "This project is a technological turning point for South Korea to secure global competitiveness in medical AI and science as a whole," and expressed his ambition to usher in an era of 'high-reliability, high-efficiency foundation models' where AI researches alongside humans and innovates clinical practices. This will be a significant step that goes beyond mere technological development, changing the paradigm of future medical research.