LG AI Research opens K-Exaone 2.5 API for free
The validation model scores close to DeepSeek V4.1 Flash on several benchmarks, and LG Uplus launches a voice AI alliance with three universities.
LG AI Research announced on November 9, 2026, that it is opening the API for K-Exaone 2.5 free of charge. The model is a technology validation version built on the second-generation independent foundation model the company unveiled in August 2026, with additional training applied.
According to techm.kr, the released benchmark figures place K-Exaone 2.5 within a narrow range of two comparison models. On Automation Bench it scored 74.6, against 74.9 for DeepSeek V4.1 Flash and 71.6 for GLM 5.3 (Max). It recorded 77.3 on Claw-Eval, versus 75.8 for DeepSeek V4.1 Flash and 77.6 for GLM 5.3.
The model posted 78.0 on BrowseComp, compared with 75.8 for DeepSeek V4.1 Flash and 79.0 for GLM 5.3, and 59.2 on SWE-Bench Pro, against 58.4 for DeepSeek V4.1 Flash and 64.0 for GLM 5.3.
The head of the Exaone Lab at LG AI Research said internal analysis showed overall performance across 24 evaluation metrics improved by more than 70% compared with the model released in August. LG AI Research also said it is considering adopting NVIDIA Dynamo for its third-generation K-Exaone model.
Voice AI alliance
Separately, LG Uplus launched the K-Voice AI Alliance with Seoul National University, Yonsei University and KAIST to jointly research voice-to-voice (V2V) technology. LG Uplus said it plans to apply the resulting technology across various services and to establish evaluation standards for Korean-language voice AI.
Quick answers
Is the K-Exaone 2.5 API free?
Yes. LG AI Research announced on November 9, 2026, that it is opening the K-Exaone 2.5 API free of charge.
What is K-Exaone 2.5 based on?
It is based on the second-generation independent foundation model LG unveiled in August 2026 and is a technology validation model with additional training applied.
Who is in the K-Voice AI Alliance?
LG Uplus launched it with Seoul National University, Yonsei University and KAIST to research voice-to-voice technology.