The US and China are clearly at the forefront of the AI race, but an unexpected nation has now established itself as the number three.
South Korea has emerged as the clear third-place contender in global artificial intelligence development ahead of France, a post by Artificial Intelligence says. South Korea leveraged a strategic government initiative and the technical prowess of its major conglomerates to produce multiple AI models that now compete at the frontier of language model intelligence.
The nation’s rapid ascent represents a departure from the US-China duopoly that has dominated AI development. According to Artificial Analysis’ Intelligence Index, South Korea now has multiple models scoring above 20 on their standardized evaluation, with LG’s K-EXAONE reaching an impressive score of 32—placing it alongside models from established American and Chinese AI labs.

The Korean National Sovereign AI Initiative: A Government-Backed Competition
A key driver of this momentum is the Korean National Sovereign AI Initiative, a government-backed, nationwide competition that incentivizes domestic model development through a multi-stage elimination process. The initiative shortlists national champions, with winners receiving direct government funding and guaranteed access to large-scale GPU capacity.
The program operates as a competitive funnel designed to concentrate resources on the most promising contenders. In August 2025, five organizations were selected: Naver, SK Telecom, LG Group, Upstage, and NC AI. In the most recent round announced last week, the field narrowed to three: LG, SK Telecom, and Upstage. A fourth finalist is expected to be selected in the coming months as the evaluation process continues.
This approach differs markedly from the venture capital-driven model prevalent in the United States or the state-directed approach of China. By creating competition among domestic players while guaranteeing infrastructure access to winners, South Korea has found a middle path that leverages both market dynamics and strategic government support.
From Outsider to Top 10: LG’s Breakthrough
South Korea’s AI ambitions first gained international attention last year when LG’s Exaone 4.0 became the only model from outside the US or China to break into Artificial Analysis’ top 10. That achievement signaled that South Korean AI labs had reached a level of technical capability previously reserved for Silicon Valley giants and Chinese tech companies.
Since then, the ecosystem has only strengthened. LG’s latest model, K-EXAONE, now scores 32 on the Intelligence Index and stands as the current leader in the Korean AI race. This 236-billion parameter open-weights model performs strongly across various intelligence evaluations, from scientific reasoning and instruction following to agentic coding. The model demonstrates frontier-level capabilities, though it exhibits high verbosity, using 100 million tokens to run the Artificial Analysis evaluation suite.
Top South Korean AI Models
Beyond LG, several other Korean companies have released competitive models, each with distinct strengths:
Upstage’s Solar Open, another shortlisted model in the national initiative, is a 100-billion parameter open-weights model scoring 21 on the Intelligence Index. It performs well in instruction following and has a lower hallucination rate compared to peer Korean models—a crucial advantage for enterprise applications.
Naver’s HyperCLOVA X SEED Think takes a different approach with a smaller 32-billion parameter reasoning model that scores 24 on the Intelligence Index. It demonstrates strong performance on agentic tool-use workflows and scores highly in the Global MMLU Lite multilingual index for Korean, highlighting its potential usefulness in a primarily Korean language environment.
Korea Telecom’s Mi:dm K 2.5 Pro represents yet another strategic direction: a proprietary reasoning model scoring 23 on the Intelligence Index with strong performance in agentic tool-use. Unlike most Korean models, Mi:dm K 2.5 Pro has no publicly available endpoint. Instead, Korea Telecom primarily intends to package this model into product offerings and use it to serve KT’s existing business clients.
Motif’s Motif-2-12.7B demonstrates that frontier-level performance is achievable even at smaller model sizes. This compact open-weights model scores 24 on the Intelligence Index and performs well in long-context reasoning and knowledge, though it is highly token-intensive, using 120 million tokens to run the Artificial Analysis evaluation suite.
Open Weights as a Strategic Advantage
A notable pattern among Korean AI models is the prevalence of open-weights releases. While American labs have largely moved toward closed, proprietary models, and Chinese labs operate under a mixed approach, Korean developers have embraced openness as a competitive strategy.
This approach offers several advantages. Open-weights models allow for greater customization and integration into existing products—crucial for conglomerates like LG that plan to embed AI across their consumer electronics ecosystem. They also build goodwill with the international AI research community and allow Korean labs to benefit from external contributions and improvements.
The strategy also aligns with South Korea’s position as a technology adopter and integrator. By making models open, Korean labs can accelerate adoption across industries while building a reputation for transparency and collaboration that distinguishes them from more secretive competitors.
South Korea’s Conglomerate Advantage
South Korea’s success in AI is inseparable from the structure of its economy, dominated by large conglomerates known as chaebols. Companies like LG, Samsung, SK, and Naver have the resources, technical talent, and long-term strategic vision to invest heavily in AI research without the pressure for immediate returns that constrains many startups.
LG’s AI Research Lab exemplifies this approach. Launched in December 2020, the lab serves as the AI research hub for the entire LG Group. By 2023, it had fostered 1,000 AI experts within LG through custom education programs and unique personnel compensation systems. The lab actively collaborates with prestigious universities including Seoul National University and the University of Toronto, and maintains research facilities in Seoul, Santa Clara, and Toronto.
For conglomerates like LG, AI model development serves multiple strategic purposes. These companies plan to integrate AI across their product ecosystems—from televisions to refrigerators to smartphones. Developing proprietary models gives them control over a critical technology stack and reduces dependence on third-party providers whose interests may not align with their own.
Implications for the Global AI Race
South Korea’s emergence as the number three AI nation carries several implications for the broader competitive landscape. It demonstrates that frontier AI development is not limited to the world’s two largest economies. With the right combination of government support, technical talent, and corporate commitment, mid-sized economies can compete at the highest levels. Also, it introduces a new model for AI development that blends government coordination with corporate competition. The Korean National Sovereign AI Initiative creates a structured pathway from research to deployment while maintaining competitive dynamics that drive innovation. It additionally signals that open-weights models remain a viable strategy for competing with closed, proprietary systems. While companies like OpenAI and Anthropic have achieved remarkable results with tightly controlled models, Korean labs are demonstrating that openness can coexist with frontier performance.
South Korea’s success may inspire other technologically advanced nations to pursue similar strategies. Countries like Japan, Germany, France, India and the United Kingdom all have strong technical capabilities and could potentially replicate elements of Korea’s approach. As the AI race continues to evolve, South Korea has established itself as more than a spectator. With multiple frontier-class models, a coherent national strategy, and the backing of some of the world’s most sophisticated technology conglomerates, the nation has secured its position as a major player in the technology that will likely define the coming decades.