LG AI Research has captured the top spot in a global industrial artificial intelligence evaluation, outperforming major U.S. and Chinese tech competitors with its proprietary EXAONE foundation models. Announced on July 7, the achievement highlights a strategic industry shift away from general-purpose chatbots toward specialized enterprise AI.
EXAONE Outscores Google and Microsoft on Tabular Benchmarks
According to LG AI Research, its tabular data prediction model, EXAONE Tabular, secured a comprehensive first place with 1,760 points in global AI performance evaluations.
The score outpaced Google’s TapFM, which scored 1,749 points, and Microsoft’s LightGBM, which recorded 1,414 points. Meanwhile, the company’s time-series prediction model, EXAONE Forecast, took first place in the zero-shot category, beating models from Google and Alibaba.
The zero-shot evaluation measures how accurately an AI predicts unfamiliar data fields without requiring additional training, according to LG AI Research.
Handling Structured Data Across Multiple Industries
EXAONE Tabular analyzes structured data found across multiple industries to handle manufacturing anomaly detection, quality judgment, risk prediction, and customer churn forecasting.
EXAONE Forecast has ingested two trillion pieces of real-world and synthetic time-series data spanning internet, energy, economic, healthcare, and transportation sectors. This vast ingestion allows the model to combine varied variables to forecast future operational scenarios.
Targeting Enterprise Niches Over General Chatbots
The performance reflects LG Group’s targeted business strategy. By bypassing direct competition with general-purpose models like OpenAI’s ChatGPT and Google’s Gemini, LG AI Research focused on enterprise market niches.
This approach allows the organization to build operational synergies with manufacturing-heavy affiliates like LG Chem, LG Energy Solution, and LG Display. A prime example includes collaboration with these affiliates to build AI models that forecast raw material prices, including lithium.
Expanding Into Healthcare, Finance, and Robotics
Building on these benchmark successes, LG AI Research is expanding its commercial footprint across specialized sectors. In 2024, the group unveiled EXAONE Path, a next-generation precision medical AI model.
The organization is currently developing an oncology agentic AI designed to assist medical professionals in cancer diagnosis and clinical decision-making. Furthermore, the company launched a stock prediction AI service last month in collaboration with Koscom and the London Stock Exchange Group (LSEG) to provide ticker-specific forecast scores and analytical commentary. Development is also underway on a robotic AI model that enables vision-based control for automated industrial tasks.
The Shift Toward Industrial-Spec AI
“Global big tech interest is rapidly moving from general language models to industrial-spec AI foundation models that solve real-world industry problems,” said Lim Woo-hyung, co-director of LG AI Research, pointing out that accumulated data from manufacturing and other physical sectors serves as a core competitive differentiator for domestic AI development.
Related reading