Overview
- South Korea named LG AI Research, SK Telecom and Upstage as the teams advancing after the first-round evaluation of its foundation model program.
- LG received the top overall mark at 90.2 points, leading benchmark tests, expert assessments and user evaluations, according to the ministry.
- Naver Cloud and NC AI were eliminated over originality concerns, with officials citing reliance on overseas pre-trained weights and references to elements of the Qwen model.
- The ministry will hold an additional selection to add a fourth contender, and eliminated teams can reapply, though Naver says it is not considering doing so.
- The project, managed by NIPA and the TTA, offers high-performance computing, curated datasets and a formal K-AI designation, with final team selections targeted by year-end.