As generative AI spreads to edge devices and industrial sites, highlighting the importance of model lightweighting and optimization capabilities in on-device environments, Nota, an AI model optimization company, held a related competition at a major domestic academic conference to expand its engagement with the developer and research ecosystem.

Nota (CEO Chae Myeong-su) announced on the 29th that it held the 'On-Device AI Optimization Contest with Netspresso' at the Korea Computer Conference (KCC 2026) held in Jeju.

This competition was based on Nota's AI optimization platform NetsPresso® and set the task of improving the execution efficiency of AI models in an environment with limited hardware resources.

The Chung-Ang University 'Eco-Pronuner' team took the championship.

This team proposed an optimization method for running LLM on small devices, reducing response latency per token by about 6.7 times and first response generation time by about 4.0 times, and reducing memory usage by about 85%.

Model quality metrics maintained existing levels, confirming the balance between speed, memory, and quality.

The Grand Prize went to Sungkyunkwan University.Team Iris won the award, while the Excellence Award went to two teams, DKE and Gamjabat.

In addition to the competition, Nota operated a recruitment booth at the KCC 2026 venue and delivered special lectures to introduce directions for the industrial application of AI optimization technology.

Kim Tae-ho, CTO and co-founder of Nota, who participated as a judge, said, “Participants used Netspresso to directly experience the entire process of AI optimization and achieved significant performance improvements in actual execution environments,” adding, “We will continue to provide support to enable easier and more efficient optimization of AI models in various hardware environments.”