Language
Change font
A
A
A
A
Share
TOP
Most read
Language
Change font
A
A
A
A
Share
The AI model developed by LG was recognized for the world’s best performance in the evaluation of global specialized institutions.
LG AI Research Institute announced on the 30th that its AI model ‘EXAONE 4.0’ ranked 11th globally and 1st among Korean models in the Intelligence Index evaluation of ‘Artificial Analysis’, a global AI performance analysis agency.
Artificial Analysis compiles seven high-level performance evaluation indicators, including ▲ Inference and Knowledge Competency (MMLU-Pro, Humanity’s Last Exam) ▲ Expert-level Science Problem Solving Ability (GPQA Diamond), ▲ LiveCodeBench, SciCode ▲ Mathematics Problem Solving Ability (AIME 2024, MATH-500).
Exemite 4.0 ranked 7th in coding and 10th in math, achieving “top 10.” The overall ranking also ranked 11th, with a score of 64 along with the top-level model of the U.S. “Claude 4 Opus.”
In particular, Exemployee 4.0 ranked fourth in the overall open weight (public) model rankings, following Q13, Dipsyck R1, and GLM-4.5, while LG AI Research Institute ranked eighth in terms of AI model development companies.
Exemite 4.0 is a hybrid AI model that combines a ‘large-scale language model (LLM)’, which is strong in understanding and generating natural language, and a ‘reason AI model’ that can hypothesize and verify itself. Exemplar 4.0 was released in two forms: a 32B (32 billion parameters) expert model and a 1.2B (1.2 billion parameters) on-device model.
On the 15th, LG AI Research Institute unveiled Exemite 4.0 as an open weight model on Hugging Face, a global open-source AI platform, so that it can be used for research, academic and educational purposes.
As a result, the 32B model achieved a record of exceeding 500,000 downloads in two weeks of release, the shortest period among domestic AI models, and currently accumulated more than 550,000 downloads.
Exemployee 4.0 was listed on the “Notable AI Model” list selected by Epoch AI, a non-profit AI research institute in the United States, along with the previously released “Exemployee Deep,” Korea’s first inference AI model. Epoch AI’s list of notable AI models is used as data for comparing AI competitiveness by country and by company in AI reports published annually by Stanford University in the United States.
LG AI Research Institute’s “Exemployee 3.5” released in December last year is also the only AI model in Korea to be included in the AI report published by Stanford University in the United States, and has announced Korea’s AI capabilities to the world.
Lee Hong-rak, co-researcher of LG AI Research Institute, said, “This result proves that LG’s Exemite has the capacity and potential to compete with the world’s best frontier AI models,” and emphasized, “We will continue our challenge to develop the world’s best AI models to achieve three global AI powerhouses.”
Meanwhile, LG CEO Koo Kwang-mo emphasized in his New Year’s address this year, “A lot of the seeds of innovation that are sprouting throughout the group, including AI and smart solutions that increase the value of customers’ time and enable them to unleash infinite potential, bio that creates a healthy life and a clean earth, and clean tech, will be a welcome value to make future customers smile.”
2026-06-12 11:11:31
2026-06-12 13:41:31
2026-06-12 10:21:18
2026-06-11 18:02:14
2026-06-12 13:20:19
2026-06-12 15:13:02
2026-06-12 15:34:01
2026-06-12 17:53:28
2026-06-12 14:50:06
2026-06-12 15:02:25
※ This article was translated using AI technology for reader convenience.
Maeil Business(MK) provides these translations “as they are” and makes no warranties of any kind, either explicitly or implicitly, regarding accuracy, reliability and marketability, suitability for a particular purpose, etc. of translation. Please be informed that the content provided may not be translated accurately due to limitations in machine translation before using this service.
Copyright (c) 매경AX. Maeil Business News Korea & mk.co.kr, All rights reserved.
Prohibition of unauthorized reproduction, redistribution, and use of AI learning