LG AI Research Opens Up API for Its Proprietary AI Model 'EXAONE 2.5' Free of Charge
LG AI Research is opening the API for 'EXAONE 2.5,' its proprietary artificial intelligence (AI) model, free of charge. An API serves as a conduit that allows developers to access and use a model's capabilities, meaning domestic developers and companies can now test the model and apply it to their own services or app development at no additional cost. LG AI Research made the announcement on the 9th at 'NVIDIA AI Day Seoul 2026,' held at COEX in Seoul.
EXAONE 2.5 is a technical validation model built on the second-generation iteration of the proprietary AI foundation model released in August, with additional training applied. It focuses on enhancing performance and usability in the field of agentic AI. Benchmarks were conducted against major AI models including China's DeepSeek V4.1 Flash and GLM 5.3 (max), centered on four metrics evaluating agentic AI capabilities. Kim Lee-reum, head of EXAONE Lab at LG AI Research, said, "Internal analysis showed that overall performance across 24 evaluation criteria improved by more than 70% compared to the model released in August."
Lim Woo-hyung, co-head of research at LG AI Research, said, "We will concentrate our research capabilities on completing a frontier-grade EXAONE model by January of next year."
