South Korea’s LG AI Research opened free API access to K-EXAONE 2.5 for developers, companies and research institutions. An API lets applications send requests to the model. LG describes this as a technology-validation version prepared through additional training of its August release. Intended uses include generating code, debugging, restructuring existing code and writing tests, alongside reasoning and workflow automation.
South Korea’s LG AI Research opened free access to K-EXAONE 2.5 on November 9 through an application programming interface, or API, which lets software send requests to a model. Developers, startups, companies and research institutions can obtain an access key and use the hosted service for development and research without a separate charge. LG announced the opening at NVIDIA AI Day Seoul at the COEX venue in Seoul.[1], [2]
K-EXAONE 2.5 builds on the second version of LG’s own foundation model, unveiled in August. LG describes it as a technology-validation model prepared through additional training, before development of the third version. The institute says this work explores improvements to the existing model’s training methods. Its stated purpose for free access is to let users try the Korean model directly and apply it to their own services.[1], [2]
LG lists reasoning, coding assistance and workflow automation among the model’s intended uses. Its announcement identifies code generation and debugging, as well as refactoring, which means reorganising existing code, and writing test code to check software behaviour. These are the development tasks LG is inviting users to explore through the service. The institute is also considering NVIDIA Dynamo technology for inference processing, the work of running a model to generate responses, and service scalability.[1], [2]
