LG's Proprietary Industrial Data Tops Google, Alibaba in AI Benchmarks

LG AI Research announced Friday that its EXAONE Tabular and EXAONE Forecast models have claimed top positions on two industrial AI benchmarks. EXAONE Tabular reached an ELO score of 1,760 on TabArena, the leading benchmark for structured-data AI, outpacing Google's TabFM model at 1,749, according to the Korea Herald. EXAONE Forecast simultaneously claimed first place in the zero-shot category on GIFT-Eval, the Salesforce-developed time-series forecasting leaderboard, besting models from Google and Alibaba, according to the Seoul Economic Daily. Both benchmarks use real industrial datasets from energy, finance, healthcare, transportation, and manufacturing. TabArena, launched at NeurIPS 2025 by Amazon Web Services and the University of Freiburg, is a continuously maintained leaderboard for tabular foundation models. Google submitted its TabFM model on June 30, 2026, and held the top spot until LG's submission. EXAONE Tabular's score places it first overall in categorical data prediction, covering binary and multi-class classification, and performs with minimal input data. LG AI Research said it plans proof-of-concept trials in manufacturing, bio-healthcare, and financial verticals in the second half of 2026. GIFT-Eval encompasses 23 datasets, over 144,000 individual time series, and 177 million data points across seven industrial domains and 10 temporal frequencies. The benchmark's primary metric for foundation models is zero-shot performance, or the ability to forecast data from a domain never seen during training without fine-tuning.