It has been assessed that South Korea has secured a "solid third place globally" in the international AI competition, following the United States and China. Analysts note that the depth and vitality of South Korea's AI ecosystem have grown due to domestic talent, public investment, and the government's "Independent AI Foundation Model" (Dokpa-mo) project, which has spurred the development of competitive models.
On the 26th, Artificial Analysis, an institutional investor evaluating AI models, stated that "South Korea has once again established a solid third-place position in the global AI competition, behind the United States and China." It further assessed that "the depth and vitality of South Korea's AI ecosystem are supported by strong domestic talent, public investment, and incentives generated by the Dokpa-mo project."
Artificial Analysis's evaluation was also utilized in the second assessment of the Dokpa-mo project. The second assessment comprised 40 points for benchmarks, 35 points for expert reviews, and 25 points for user feedback, with Artificial Analysis's results reflected in the benchmark evaluation.
In fact, multiple domestic companies are releasing models that demonstrate competitiveness in global evaluations. Four models from Motif Technology, Upstage, SK Telecom, and LG AI Research recorded scores of 30 or higher on Artificial Analysis's "Intelligence Index."
Among domestic models, Motif Technology's "Motif 3" received the highest score of 47 points. It is an open-weight model with a total parameter count of 314 billion and active parameters totaling 13 billion. Upstage's "Solar Pro 4" achieved a score of 42. These two models recorded the highest scores on the Intelligence Index, excluding models developed in the United States and China.
SK Telecom's "A.X K2" scored 35 points, while LG AI Research's "K-Exaone 2.0" scored 31 points. Their total parameter counts are 692 billion and 750 billion, respectively. However, in individual model performance, the United States remains ahead; for instance, Anthropic's "Claude Opus 4.6" from the U.S. achieved a score of 63. South Korea's "third-place" assessment emphasizes not that a single model has surpassed top-tier U.S. and Chinese models, but rather that multiple competitive model development companies have emerged, thickening the layers of the ecosystem.
Ha Jeong-woo, former Senior Secretary for AI Future Planning at the Presidential Office, stated on X (formerly Twitter) that day: "The globally most widely recognized Artificial Analysis has officially recognized South Korea as the No. 3 nation in the AI race." He further evaluated this as a joint achievement by domestic Dokpa-mo companies possessing national AI strategies and innovation capabilities. He also emphasized the need to concentrate support and investment on companies with innovation achievements, capabilities, and determination, regardless of their size.
Artificial Analysis noted that Trillion Labs and KT are also developing their own AI model groups, assessing that South Korea's AI model ecosystem has become deeper and more competitive compared to a year ago. With the number of competitive model development companies increasing, laying the foundation for "third place globally," the challenge ahead is not only quantitative growth of the ecosystem but how quickly the performance gap with top-tier U.S. and Chinese models can be narrowed.
in TECH
