A representative from Naver Cloud is introducing the AI capabilities of Team Naver to international AI researchers participating in ‘NeurIPS 2025.’ /Naver Cloud Team Naver announced on the 9th that it has proven its research capabilities and competitiveness in AI technology by having 10 papers accepted at the world-renowned AI conference ‘NeurIPS 2025.’
At this conference, Team Naver shared its AI full-stack strategy under the theme ‘From Research to Reality,’ featuring its hyper-scale AI model ‘HyperCLOVA X’ and Naver’s industry-specific AI services.
The papers accepted this time cover innovations in three technical areas: efficient operation of hyper-scale AI, enhancement of safety in generative AI, and expansion of AI into the physical world. Particularly notable were technologies that can save up to four times the memory and double the speed without performance reduction during LLM’s context processing, as well as technologies that improve the quality of AI-generated content and ensure safe control.
Additionally, research extending AI to the real world was also presented. Model structures that help robots remember changing spaces and find paths demonstrate the practical applicability of AI technology. Team Naver built a network with over 29,000 AI researchers at this conference and shared its technological vision.
A Team Naver representative stated, “Having 10 papers accepted at NeurIPS is evidence that our research capabilities are being recognized globally,” and added, “We will continue to strengthen our research capabilities to realize AI technology in the industry.”