LG AI Research unveiled ‘K-Exaone 2.0’, the second model of the Ministry of Science and ICT-led ‘Independent AI Foundation Model Project’, on July 31. With 750 billion parameters, the largest in South Korea, the model is over three times larger than the first version (236 billion parameters). LG AI Research made the model openly available on the Hugging Face community platform, allowing anyone to use it for commercial purposes. ‘K-Exaone 2.0’ scored an average of 70.1 points across 24 benchmark evaluation metrics, a 10% improvement over the first model’s 63.3 points. Notably, it showed a 30% average performance increase in three key metrics for coding and agentic coding. In evaluations for long-context understanding, agentic capabilities, and instruction-following, it matched or outperformed Zhipu’s ‘GLM-5.1’ and Alibaba’s ‘Qwen 3.5’. Supported languages expanded from six—Korean, English, Spanish, German, Japanese, and Vietnamese—to 10, adding French, Italian, Portuguese, and Polish. Im Woo-hyung, co-head of research at LG AI Research, stated, “K-Exaone 2.0 is significant because domestic researchers independently completed the entire process, from designing a 750B-class model to data training, distributed learning, and inference environment setup,” adding, “