Naver Cloud Inference AI Model Listed on Global Benchmark, Ranks Second in Korea

Scored 44 Points on Artificial Analysis Intelligence Index
Outperformed by LG Exaone and Upstage

Naver Cloud's newly released high-performance inference artificial intelligence (AI) model has been newly listed on a global AI evaluation index. This model ranked among the top domestic models.

According to the global AI performance evaluation organization Artificial Analysis (AA) on December 30, Naver Cloud's high-performance inference model, "HyperCLOVA X Seed Think (32B)," which was unveiled the previous day, has been newly included in the "Intelligence Index" calculated by this organization.

The high-performance inference model of 'HyperCLOVA X,' listed on the 'Intelligence Index' by the global AI performance evaluation organization 'Artificial Analysis (AA),' is being utilized by the company...

On this index, Naver Cloud's inference model scored 44 points, which is the second-highest score among domestic AI models. Among domestic models, Motif Technologies' "Motif 12.7B" received the highest score with 45 points, one point higher.

Another domestic AI model, "Exaone 4.0 (32B)" from LG AI Research, scored 43 points, while Upstage's "Solar Pro2" model received 38 points.

Among all models, Google's "Gemini 3 Pro" and OpenAI's "GPT-5.2" both received the highest evaluation, tying at 73 points.

Additionally, Naver Cloud's inference model achieved the highest score among domestic AI models, recording 87% in the category evaluating agent tool usage ability by simulating customer support scenarios for telecommunications companies. This is seen as proof of its potential as an "Omni-modal" agent. The omni-modal concept is a next-generation AI technology that is an expanded version of the existing "multi-modal" approach.

Previously, Naver Cloud unveiled two new AI models, including this inference model. In addition to the high-performance inference model, the native omni model "HyperCLOVA X Seed 8B Omni" was also released. Native omni-modal refers to a model that learns text, images, and voice simultaneously from the beginning within a single model.