Korea Ranks Third Globally in Notable AI Model Releases, Excels in 'Physical AI'
Translated from Korean, summarized and contextualized by DistantNews.
At a glance
- South Korea ranks third globally in the release of notable AI models and 18th in AI adoption, with a significant increase in adoption rates.
- The country leads in industrial robot installations per capita ('robot density') and shows strength in 'physical AI,' applying AI to factories and robots.
- Despite strong AI patent applications, a gap remains in the absolute performance of Korean AI models compared to top US and Chinese counterparts.
South Korea has emerged as a significant player in the global artificial intelligence landscape, ranking third worldwide for the release of notable AI models and 18th in AI adoption rates, according to the "2026 AI Index" report by Stanford University's Human-Centered Artificial Intelligence (HAI) Institute. The report, analyzed by POSCO Management Research Institute, highlights a notable improvement in Korea's AI adoption, climbing seven spots from the previous year. The increase in adoption rates was the highest globally, with a 4.8 percentage point rise.
Korean industry has entered a 'transition momentum' phase where AI absorption is accelerating, and Korea's strengths are particularly evident in the 'physical AI' field, which applies artificial intelligence to factories and robots.
This surge in AI adoption is particularly evident in the industrial sector, where the installation of industrial robots has reached 30,600 units, placing Korea fourth globally behind China, Japan, and the United States. More impressively, Korea maintains its top position in 'robot density,' measured by the number of robots per 10,000 workers, with 1,012 units. The report identifies 'physical AI', the application of AI in factories and robotics, as a key area of strength for the Korean industry, which is described as being in a "transition momentum" phase of AI absorption.
In terms of technological innovation, Korea released five notable AI models in large language models (LLMs) in the latter half of last year, positioning it third globally behind the US (50 models) and China (30 models). These Korean LLMs, developed by companies like LG AI Research, Naver, SK Telecom, Upstage, and NCSoft, are characterized by their medium scale (around 30 billion parameters). This strategic focus on mid-sized models differentiates them from the massive, trillion-parameter models from the US and China, emphasizing "cost-efficiency" and "Korean language specialization."
Korean large language models are concentrated in the mid-sized category with about 30 billion parameters, showing a strategy of differentiation in terms of 'cost-efficiency' and 'Korean language specialization' compared to the top US and Chinese models with hundreds of billions to trillions of parameters.
Korea also leads the world in AI patent applications per capita, with 14.31 applications per 100,000 people for two consecutive years, indicating the most active innovation relative to population size. Luxembourg, China, and the US follow. However, the report cautions that patent numbers do not guarantee commercial success or model quality. It points out a significant performance gap between Korea's indigenous models and the top-tier models from the US and China. The overall AI market in Korea was valued at approximately $9.06 billion last year, representing 1.6% of the global market.
Although the number of patent applications is high, it does not directly guarantee commercial success or model quality. Compared to world-class models, the absolute performance of Korea's own models still shows a significant gap compared to the top models in the US and China.
Originally published by Hankyoreh in Korean. Translated, summarized, and contextualized by our editorial team with added local perspective. Read our editorial standards.