Twelve Labs, founded in 2020 by Jae Lee, builds AI models and infrastructure specifically for understanding, searching, and analyzing video content, rather than treating video as a series of still frames the way most general-purpose vision models do.
Its Marengo 3.0 model reads every sound, word, and motion on screen across time, and that video-native approach drew a $100 million Series B in July 2026, co-led by NEA and NAVER Ventures with Amazon and Nvidia's venture arms among the participants, bringing cumulative funding past $200 million.
Pricing runs through API access with usage-based rates for video processing and search, with details available through Twelve Labs' developer platform. For a developer who needs to search and analyze video content by what actually happens on screen, rather than relying on manual tagging or frame-by-frame image analysis, Twelve Labs' video-native models address that directly.
