Search results for Solar Open2
AI & Enterprise
Dokpamo phase 2 to assess agentic AI, allow tool calls in expert review
As competition among large language models shifts from simple question-and-answer to agentic AI that uses external tools and data, attention is growing on how the government’s “independent AI foundation model” project evaluates such capabilities. The second-phase Dokpamo evaluation keeps benchmark, expert and user reviews, but splits user testing into professional users and 200 members of the public. The science ministry says it differentiates core model performance from agentic capability by varying tool-use rules.
AI & Enterprise
Dokpamo participants bet on agentic AI ahead of August second-round review
Companies taking part in Korea’s independent AI foundation model project, known as Dokpamo, are stepping up efforts to build agentic AI capabilities ahead of a second-stage evaluation in August. Developers are focusing on orchestration skills such as tool calling as large language model use expands beyond simple Q&A to real work tasks. LG AI Research, Upstage, SK Telecom and Motif Technologies each say they are strengthening agent performance in their models for the review.
AI & Enterprise
Upstage CEO Kim Sung-hoon says Fable5 foreign access ban highlights need for Korean sovereign AI
Upstage, a South Korean AI startup, said it aims to compete in global technology rivalry with sovereign AI. CEO Kim Sung-hoon said AI has become a national strategic asset and countries can cut off access. He cited a recent U.S. move to block foreigners from accessing Anthropic’s Fable5 and Mythos5. Upstage said its open-source Solar Open2 preview scored 44.4 on AAII 4.0 and is comparable to leading models.