South Korea launches medical and bio-specialized AI foundation model projects

South Korea is accelerating the development of large-scale artificial intelligence (AI) foundation models tailored for medical and biotechnology applications. The Ministry of Science and ICT announced on October 31 that it has selected the Lunit consortium and the KAIST consortium to lead national initiatives to develop advanced medical and bio AI models.

As part of the program, each consortium will receive 256 units of Nvidia’s latest Blackwell (B200) GPUs by early September next year, providing the high-performance computing infrastructure required to train global-scale models.

Lunit, a leading medical AI company, is forming a multidisciplinary consortium that includes Trillion Labs, Kakao Healthcare, SK Biopharmaceuticals, Eisen Science, Standigm, Rebellions, KAIST, Seoul National University, NHIS Ilsan Hospital, Yongin Severance Hospital, and Kyung Hee University Hospital at Gangdong.

The consortium aims to develop a “full-cycle medical science AI foundation model” that integrates clinical evidence, diagnostic workflows, and real-world medical datasets. The model is expected to scale up to 32 billion parameters, surpassing existing global medical AI models, and will be developed from scratch rather than via tuning or adaptation.

Once developed, the model will be released as open source and deployed on Kakao Healthcare’s platform for public medical services. The consortium also plans to expand the model to a 1-trillion-parameter scale over time through extended private investment.

Meanwhile, the KAIST consortium will focus on next-generation bio foundation models under the project name “K-Fold.”

Participants include Merck, HITS (AI drug discovery), AtoLab, the Korea Pharmaceutical and Bio-Pharma Manufacturers Association, and the Korea BIO Association.

Unlike conventional multiple sequence alignment (MSA)-based prediction, the K-Fold model will learn causal physical and chemical interaction patterns within proteins and molecular systems.

The project will produce both a 7B-scale foundation model and a 2B-scale lightweight SaaS model, and will also be released as open source to enable widespread research and commercial adoption.

According to the Ministry, integrating Korea’s strengths in medical science, clinical research, and pharmaceutical innovation with large-scale AI is expected to accelerate advancements in diagnostics, new drug development, and bio research.

In parallel, the government plans to establish a secure medical data-sharing ecosystem and a revenue-sharing framework to ensure fair distribution of innovation benefits.

A ministry official stated, “Korea has strong foundational capabilities in medical and bio research.

By integrating AI at a national level, we aim to create global competitiveness and drive a new wave of healthcare innovation.”