LG Uplus Partners with OptAI to Develop AI Token Optimization Technology

On October 2, LG Uplus announced that it will partner with OptAI, a company specializing in AI model optimization, to develop token optimization technology that aims to improve AI operational efficiency.

Lee Sangyeop, CTO of LG Uplus (Executive Vice President, left), and Lee Jaeho, CEO of Opt AI, pose for a commemorative photo related to token optimization collaboration. LG Uplus.

Token optimization is a technology that enables AI models to handle a greater number of requests with the same resources either by making the models lighter or streamlining computational processes.

LG Uplus will be responsible for verifying and applying the technology, leveraging its practical experience in running AI services. OptAI will focus on research and development of technologies that allow AI models to run lighter and faster.

Previously, the two companies applied lightweight technology to a small language model (sLM) based on 'Exaone', making it operable on the neural processing unit (NPU), an AI-dedicated semiconductor in smartphones. While maintaining performance similar to the existing central processing unit (CPU) approach, they managed to reduce power consumption by 78% and model size by 82%.

The companies plan to expand their joint research scope to include server graphics processing unit (GPU) environments. In particular, by improving the computational process that generates responses from AI models to better fit real service environments, they have increased the number of tokens that can be processed on a single GPU by up to four times compared to previous capabilities.

Lee Sangyeop, CTO of LG Uplus, said, "We will expand our research on token optimization technology in collaboration with OptAI and other partners to deliver results that can be applied to real-world services."

This content was produced with the assistance of AI translation services.