Naver Cloud's HyperCLOVA X SEED Think (32B) reasoning model scored 44 points Tuesday on the Intelligence Index compiled by Artificial Analysis (AA), a global AI performance evaluation organization.
The language models are tested by AA through 10 evaluation datasets, testing each models' capabilities across reasoning, knowledge, math and programming.
The model trailed only Motif Technologies' Motif-2-12.7B, which scored 45 points. Other Korean models included LG AI Research's EXAONE 4.0 32B with 43 points and Upstage AI's Solar Pro2 at 38 points. Google's Gemini3 Pro and OpenAI's ChatGPT 5.0 topped the index with 73 points each.
Naver's model also achieved 87 percent in an evaluation that assessed agent tool usage capabilities through a simulated telecommunications customer support scenario, the highest score among Korean AI models.
Industry observers said the results demonstrate the model's potential as an omni-modal agent, a next-generation AI technology that extends beyond conventional multi-modal capabilities.
Naver Cloud unveiled two omni-modal models on Monday that can simultaneously understand and generate text, images and voice from the outset, including the reasoning model HyperCLOVA X SEED Think and an upgraded version of its HyperCLOVA X platform called the Native Omni Model.
Copyright ⓒ Aju Press All rights reserved.