LG AI Research's Exaone models top global AI benchmarks

LG AI Research announced that its Exaone foundation models for tabular data (Exaone Tabular) and time‑series forecasting (Exaone Forecast) achieved the highest scores on worldwide AI benchmark leaderboards, surpassing competing models from Google and Alibaba. Exaone Tabular recorded an ELO rating of 1,760, edging out Google’s TabFM (1,749), while Exaone Forecast secured first place in zero‑shot forecasting across sectors such as finance, energy, manufacturing and healthcare.

The models are already deployed within LG affiliates to predict product demand at LG Electronics and raw‑material prices at LG Energy Solution, and are being used with partners Koscom and the London Stock Exchange Group to analyse South Korean and U.S. stock markets. LG AI Research plans to broaden the Exaone lineup with a pathology foundation model (Exaone Path) for cancer diagnosis and a robotics foundation model for visual‑information‑driven robot control. Co‑head Lim Woo‑hyung said the results demonstrate that industrial data can give Korea a differentiated AI advantage as global tech firms shift toward industry‑specific solutions.