Companies & labs

LG AI Research

LG Group's AI research organisation that develops the EXAONE models

6recordsFirst record 2024Latest Jul 31, 2026

Profile

Country
South Korea
Headquarters
Magok, Gangseo-gu, Seoul (LG Science Park area)
Founded
7 December 2020
Founded by
LG Group (under LG Management Development Institute, wholly owned by LG Corp.)
Type
Unlisted research organisation within LG Management Development Institute

LG AI Research is the artificial intelligence (AI) research organisation of South Korea's LG Group. It was launched on 7 December 2020 as a group-wide organisation dedicated to AI. It is not a separately listed company: it sits under LG Management Development Institute, which is wholly owned by LG Corp., the group's holding company (a company that owns shares in the group's affiliates and steers the group). Sixteen affiliates, including LG Electronics and LG Chem, took part at its launch. In May 2025 it moved its headquarters from Yeouido in Seoul to the LG Science Park area in Magok, Gangseo-gu, Seoul. As of October 2026, it is led by two co-heads, Lee Honglak and Lim Woo-hyung.

LG AI Research studies the latest core AI technologies (key technologies that other technologies build on) for the whole group and uses AI to solve problems that affiliates find hard to crack. Typical examples are predicting battery life and finding candidate substances for new medicines. The institute said that by September 2026 it had solved more than 100 such industrial problems. LG Corp.'s results materials show that research services provided by the AI research institute (work done on commission for a fee) are reflected in LG Management Development Institute's revenue. It also offers EXAONE-based AI specialised for particular fields, such as inspection in manufacturing processes, designing new materials and financial forecasting.

Its core technology is the EXAONE family of AI models. It unveiled its first model, EXAONE 1.0, in December 2021. EXAONE 4.0, released in July 2025, is a 'hybrid' model that combines everyday language ability with step-by-step reasoning in a single model. It came as an expert model with 32 billion parameters (the numbers an AI adjusts as it learns) and a 1.2-billion-parameter model that runs on a device without a server connection. In December 2025 it unveiled K-EXAONE (236 billion parameters) under the 'Sovereign AI Foundation Model Project' run by the Ministry of Science and ICT, and in April 2026 it announced EXAONE 4.5, which understands text and images together. In July 2026 it published K-EXAONE 2.0, with 750 billion parameters, on Hugging Face (a site for sharing AI models) under the Apache 2.0 licence (terms of use), which also allows commercial use.

Made by AIThis profile was written by AI (Claude) from the 13 sources above (official pages, filings, news reports).How it's made

Records 6

Newest first. Open one to see it with its source.

LG AI Research
LG AI Research releases K-EXAONE 2.0 (750B-A37B)

More than 3 times the 236B of phase 1. The switch to Apache 2.0 removes commercial restrictions

LG AI Research
LG AI Research releases EXAONE 4.5 33B

As LG's first open-weight vision-language model, it marks the point where Korea's open-weight route expanded beyond text

LG AI Research
LG AI Research releases EXAONE 4.0

A hybrid that integrates reasoning and agent capabilities (MCP, function calling)

LG AI Research
LG AI Research releases EXAONE Deep

LG's open-weight line specialized for math, science and coding reasoning; company-reported scores for the 32B were 94.5 on the 2025 CSAT math section and 90.0 on AIME 2024

LG AI Research
LG AI Research open-sources EXAONE 3.5

A follow-up four months after the EXAONE 3.0 7.8B release that expanded to three sizes while keeping a non-commercial license

LG AI Research
LG AI Research releases EXAONE 3.0

The first open release of an in-house LLM by a major Korean company. But its research-only license also revealed the conservatism of the Korean approach to openness