NVIDIA Brings Korea's Leading Open AI Models to Production With Dynamo Recipes

OFFICIAL Written by GPN · Nov 9, 2026, 9:45 AM ET · · Tech

NVIDIA has released deployment recipes that bring Korea's leading open AI models into production, built on its open-source Dynamo inference framework.

The recipes cover four models: A.X K2 from SK Telecom, K-EXAONE 2.0 from LG AI Research, Solar Open 2 from Upstage and Motif-3 from Motif Technologies.

As models grow beyond what a single GPU can hold, they have to be split across several GPUs, with careful coordination of request routing and communication to keep performance up. Dynamo, NVIDIA's open-source distributed inference-serving framework, handles that routing and communication for multi-node, data-center-scale deployments.

Optimized for NVIDIA B200 GPUs, the recipes deliver up to twice the performance out of the box, according to NVIDIA, taking the guesswork out of deployment and helping developers turn Korea's open models into broadly accessible AI services faster.

The release is part of a wave of Korea-focused AI announcements from NVIDIA published on November 9, including new collaborations on sovereign AI infrastructure and physical AI.

Facts gathered from

NVIDIA Blog

NVIDIA Brings Korea's Leading Open AI Models to Production With Dynamo Recipes

Facts gathered from

More on this