California: LG AI Research, the artificial intelligence lab under South Korea's LG Group, announced the launch of its new inference AI model, EXAONE Deep, marking a significant step in its pursuit of agentic AI. The unveiling occurred during the GTC 2025 event hosted by AI chip giant Nvidia Corp., which took place from Monday through Friday in California. The EXAONE Deep model has been released as an open-source model.
According to Yonhap News Agency, agentic AI employs sophisticated reasoning and iterative planning to autonomously tackle complex, multistep problems. The development of such AI systems necessitates advanced AI technologies, like inference AI, to be brought into practice. LG AI Research noted that EXAONE Deep is currently the only South Korean model capable of competing with other global inference AI models, a space presently dominated by major global tech firms with foundational AI models such as OpenAI's GPT series.
In addition to the primary EXAONE Deep-32B model, LG AI Research introduced two other models: the lightweight EXAONE Deep-7.8B and the on-device EXAONE Deep-2.4B, both of which are also available as open-source models. A technical report evaluation highlighted the main model's superior performance, awarding it top scores in mathematics, science, and coding compared to global open-source AI models from the United States, China, and other countries.
LG AI Research expressed its intention to lead the on-device AI market by further enhancing its AI models through collaboration with LG Group affiliates, including LG Electronics Inc. and LG Uplus Corp.