Moments Lab has unveiled MXT-2, the latest version of its multimodal AI indexing technology, designed to help producers quickly locate key moments in large video libraries for content creation, repurposing, and monetization. With a training dataset exceeding 1.5 billion images—more than three times larger than its predecessor—MXT-2 delivers even more detailed and accurate humanlike descriptions of video content.

New features include Custom Moments, which provide time-coded descriptions of key scenes tailored to user needs. This allows producers to pinpoint compelling moments in unscripted TV shows for use in viral shorts, trailers, or best-of compilations. Additionally, the Custom Insights feature automates thematic classification and content generation, enabling users to create titles, teaser descriptions, and even full-length articles with AI-powered text generation.

MXT-2 is now available on the Moments Lab video discovery platform and can also be integrated into existing tools, workflows, and platforms via API.