TwelveLabs Launches Groundbreaking Video AI 'Pegasus-1.2'

Twelve Labs, a leader in video AI technology, unveiled its innovative video language generation model 'Pegasus-1.2' on the 12th, ushering in a new era of video understanding. This model goes beyond simple video analysis, boasting the ability to deeply understand both visuals and audio simultaneously and perfectly convert them into text. It can process any complexity, from short clips to feature-length videos of an hour, with high accuracy. Seungjun Lee, CTO of Twelve Labs, emphasized, "Pegasus-1.2 possesses the ability to grasp the essence of video through its innovative spatiotemporal information understanding method, which will meet the demands of various industrial sectors." Particularly noteworthy is its competitive edge. Pegasus-1.2 demonstrates significantly faster response times than GPT-4o and Gemini 1.5 Pro, offering industry-leading performance at a lower cost. This is thanks to the innovative combination of efficiency and comprehension created by its advanced vision encoding strategy and sophisticated token reduction technology. Its application fields are limitless. In entertainment, it creates new value through video content classification and highlight extraction, while in education, it maximizes learning efficiency by summarizing key points and extracting essential information from lecture videos. Furthermore, in the security sector, it will strengthen our society's safety net by detecting anomalous signs through CCTV video analysis. Since its founding in 2021, Twelve Labs, which has focused on the potential of video multimodal technology, has once again proven its unique technological prowess by being named among the 'Top 50 Generative AI Startups Worldwide' by global research firm CB Insights. Pegasus-1.2 is another masterpiece born from their relentless innovation.