TLDR: LG AI Research hosted its ‘LG AI Talk Concert 2025’ in Seoul, showcasing the evolution of its EXAONE ecosystem, including advanced AI models and services for industrial applications. Highlights included the debut of EXAONE 4.0 VL, a multimodal AI model, and the unveiling of EXAONE-based enterprise AI agents like ChatEXAONE, EXAONE Data Foundry, and EXAONE On-Premise, demonstrating LG’s commitment to AI innovation and real-world application.
SEOUL, South Korea – LG AI Research convened its highly anticipated ‘LG AI Talk Concert 2025’ today at the LG Sciencepark Convergence Hall in Magok, Seoul, marking a significant milestone in the evolution of its EXAONE ecosystem. The event underscored the culmination of five years of dedicated AI development, spotlighting advanced AI models and services meticulously designed for diverse industrial applications.
Woohyung Lim, Co-President of LG AI Research, articulated the company’s ambitious future trajectory, stating, “LG AI Research will secure global competitiveness based on our proprietary foundation model, go beyond AI model development to apply it in various industrial fields, thereby achieving both versatility and specialization, and create an AI ecosystem with global partners.”
Honglak Lee, Co-President of LG AI Research and Chief Scientist of AI (CSAI), introduced the recently unveiled EXAONE 4.0, a leading hybrid AI model developed in Korea. A key highlight of the concert was the debut of EXAONE 4.0 VL (Vision Language), a sophisticated multimodal AI model engineered to comprehensively understand complex professional documents, images, and molecular structures. In performance comparisons, EXAONE 4.0 VL demonstrated superior capabilities compared to Meta’s Llama 4 Scout model. Mr. Lee emphasized that EXAONE 4.0 VL is poised to serve as the ‘eyes’ of the EXAONE model.
Further advancements in the medical field were showcased with EXAONE Path 2.0, a precision medical AI model capable of dramatically reducing disease diagnosis time from two weeks to under one minute.
Jungkyu Choi, leader of the AI Agent Group at LG AI Research, subsequently unveiled a suite of EXAONE-based enterprise AI agents that have successfully completed LG’s internal validation. These include ChatEXAONE, an AI agent designed for LG employees, which has obtained ISO certification for its ability to process even national core technology documents, indicating its potential for broader expansion.
The newly introduced ‘EXAONE Data Foundry’ was presented as a pivotal AI factory for producing high-quality data. This platform significantly streamlines data generation, enabling a single individual to accomplish in one day what would typically require 60 experts working for three months. Validation projects with LG affiliates and national institutions have shown a minimum 1,000-fold increase in data productivity and an average 20% improvement in data quality compared to existing methods.
Concluding the showcase, LG AI Research introduced ‘EXAONE On-Premise,’ a full-stack solution built entirely with proprietary Korean technology, spanning from AI semiconductors to models. This solution allows companies to establish enterprise agents in an independent environment, enabling the secure utilization of EXAONE models without external security concerns. LG AI Research also highlighted its key domestic and international partners within the EXAONE ecosystem, signaling a strategic move towards diversifying its AI B2B business model.
Also Read:
- LG CNS and Honeywell Forge Alliance to Revolutionize US Manufacturing with AI
- Oracle Revolutionizes Enterprise Productivity with Advanced AI Agent Integration
This year’s event follows LG AI Research’s continuous pursuit of innovation, including the release of EXAONE Deep in March, Korea’s first reasoning AI model, and EXAONE 4.0 in July, Korea’s first hybrid AI model combining general and reasoning capabilities.
