[Edaily Reporter Kim Hyun-ah ] Lim Woo-hyung, Co-Director of the LG Corp. AI Research Institute, stated, “It is difficult to narrow the technological gap with global Big Tech using only small models,” adding, “We will continue to take on the challenge with the goal of achieving performance equal to or better than that of our competitors at the same scale.”

LG Corp.(003550) LG AI Research’s “K-EXAONE” passed the second evaluation of the “Independent AI Foundation Model Project (DOKPAMO),” organized by the Ministry of Science and ICT, on the 18th and advanced to the third evaluation.

Aiming to develop a hyper-scale foundation model capable of competing in the global market, LG Corp. AI Research expanded the model’s scale to more than three times that of existing models and focused on securing technologies across the entire spectrum of training and inference for hyper-scale models.

In the second evaluation, the institute focused on establishing a technological foundation capable of stably training and operating ultra-large models, while simultaneously improving response quality and performance to a level that real users can perceive.

Co-Director Lim stated, “Significantly expanding the model’s scale was an inevitable strategic choice to compete on equal footing with global frontier-level models,” adding, “We will continue this challenge with the goal of achieving performance equal to or better than that of models in the same scale class.”

He continued, “While the improvement in performance compared to the first evaluation may appear somewhat modest, this is a necessary technical validation process we must undergo to make the leap toward the frontier-level AI that LG Corp. aspires to achieve,” and emphasized, “This second phase involved optimizing the training of the ultra-large model over a limited period, and based on the technical assets and know-how accumulated during this phase, we will successfully achieve our final target performance.”

LG Corp. AI Research plans to develop K-ExaOne into a frontier-level expert AI model representing South Korea, building on the technology secured during this phase. The goal is to establish a technological foundation capable of competing with leading global models, going beyond AI optimized for the domestic environment.

LG Corp. AI Research plans to hold a talk concert in about a month to provide a detailed explanation of the technologies acquired through this project and the future development direction of K-ExaOne.