Seoul: LG AI Research, the artificial intelligence (AI) lab under South Korea's LG Group, has unveiled EXAONE 4.0, marking the country's first hybrid AI model. The hybrid AI model is designed to deliver world-class performance by combining the strengths of large language models (LLMs) and reasoning AI, as stated by the company in a press release.

According to Yonhap News Agency, this launch is LG's second major AI milestone this year, following the introduction of South Korea's first reasoning AI model, EXAONE Deep, in March. Lee Jin-sik, head of the EXAONE Lab at LG AI Research, emphasized the company's commitment to research and development, aiming to establish EXAONE as Korea's leading frontier AI model and to showcase its competitiveness on the global stage.

Benchmark performance tests revealed that EXAONE 4.0 outperformed leading open-weight models from the United States, China, and France. Globally, only a few companies have introduced similar hybrid AI models. U.S.-based Anthropic and China's Alibaba have released their Claude and Qwen models, respectively, while U.S. AI firm OpenAI is reportedly developing GPT-5 as a hybrid model.

LG has made EXAONE 4.0 available as an open-weight model on Hugging Face, an international open-source AI platform, for research and educational purposes. An open-weight model refers to an AI model whose learned parameters, or weights, are publicly released, allowing others to use and adapt the model, even though its training data and architecture may remain proprietary.

In collaboration with Friendly AI, an official model distributor on Hugging Face, LG has launched a commercial application programming interface (API) service for EXAONE 4.0. The company anticipates that the API release will accelerate AI adoption across various industries.

Furthermore, LG is planning to host an AI-themed talk concert at LG Science Park in Seoul on July 22. The event will feature the showcase of EXAONE 4.0, highlight related AI technologies, and announce future research projects.