Naver to unveil HyperCLOVA X to rival global big techs this week

[Courtesy of Naver]

Amid the growing dominance of global tech giants in the realm of hyper-scale artificial intelligence (AI), South Korean platform giant Naver Corp. is poised to unveil its own Large Language Model (LLM) known as HyperCLOVA X on Thursday.

According to the information technology (IT) industry sources on Monday, Naver stands at the forefront of large-scale AI advancement in Korea. Since 2021, Naver has made significant strides in the development of its hyper-scale AI model based on proprietary supercomputing capabilities and indigenous data.

HyperCLOVA is distinguished by its capability to handle vast data volumes by utilizing a supercomputer that can execute 700 petaflops, a milestone introduced by a Korean enterprise, diverging from the convention of relying on external clouds. HyperCLOVA X is an upgrade version with enhanced performance, featuring 204 billion parameters.

In a landscape where Google’s PaLM 2 and Microsoft’s GPT-4 are vying for supremacy in the realm of ultra-large AI LLMs, the spotlight now turns to Naver’s HyperCLOVA X.

Securing domestic original technology to compete with global big techs had become a hot issue since the end of last year as the dominance of global AI leaders with GPT series accelerated.

HyperCLOVA X has 204 billion parameters, exceeding those of GPT-3 with 175 billion, and has learned 6,500 times more Korean languages than ChatGPT, which is based on GPT-3.5.

Along with HyperCLOVA X, Naver will also introduce CLOVA X, which can be used as a plug-in to call necessary functions, and Cue:, an interactive AI chatbot developed specifically for search.

The company also plans to showcase a blueprint for combining these services with existing businesses such as commerce, creatives, and finance.