Five Companies Including Naver, SK Telecom, LG, Upstage, and NC AI
to Hold Public Presentation at COEX on the 30th
Showcasing Ultra-large, Omni-modal, and Industry-specialized Models
First Evaluation Begins Immediately After Presentation... Results to Be Announced Next Month
The first presentation of the Reader AI Foundation Model Project will be held at COEX on the 30th. Image=Gemini generated
The first achievements of the government-led "Reader AI Foundation Model" project will be released on the 30th. The Ministry of Science and ICT and the National IT Industry Promotion Agency will hold the first presentation of the Reader AI Foundation Model Project at COEX Auditorium in Seoul on the 30th, where five elite teams will publicly unveil the results of the AI models they have developed over approximately four months for the first time.
The presentations will proceed in the order of Naver Cloud, NC AI, Upstage, SK Telecom, and LG AI Research. Each team will have 20 minutes to present and demonstrate their models, directly proving their technological capabilities and potential applications.
The government has outlined its vision to reduce dependence on global big tech companies and secure "sovereign AI" capabilities based on Korean proprietary technology through this project. The first presentation is significant as it marks the first opportunity to verify whether this vision can move beyond declarations and translate into tangible technological achievements.
Each consortium is highlighting its unique strengths ahead of the presentation to stand out. SK Telecom is emphasizing that it has the largest domestic model by introducing the country's first ultra-large AI model with over 500 billion parameters. Naver Cloud is showcasing the technological scalability of its omni-modal "HyperCLOVA X," which integrates text, image, voice, and video processing within a single model.
Upstage aims to expand AI services for the general public and industry-specific solutions based on its proprietary "SOLAR" model. NC AI is focusing on developing industry-specialized models for manufacturing, distribution, and robotics centered on its 20 billion-parameter model. LG AI Research is emphasizing advanced reasoning and complex generative capabilities based on its next-generation "K-Exaone."
The key points of interest at this presentation will extend beyond the absolute scale of the models to include Korean language and industry-specific performance, multimodal and reasoning capabilities, and the models' applicability to real-world services and public or industrial use. In the lobby, there will also be booths where general attendees can experience the AI models firsthand. According to the event's organizing office, advance registration has exceeded the venue's capacity, so on-site entry may be limited to a first-come, first-served basis, and the event will also be broadcast live on YouTube.
Immediately after the presentation, the first phase of evaluation will take place. The results of the first evaluation are scheduled to be released by January 15 next year, serving as a critical milestone that will determine whether each consortium can advance to the next stage and further refine their models. The extent to which each team achieves the government’s goal of "95% of the latest global models" will also be a key focus of the evaluation.
An AI industry official commented, "This presentation is the first testbed for assessing the direction and potential of a Korean foundation model," adding, "It will serve as a signal for whether the domestic AI industry can progress toward an independent ecosystem."
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
