TwelveLabs Debuts Marengo 3.0 Video AI Model on AWS Bedrock

December 1, 2025

Details

  • TwelveLabs has launched Marengo 3.0, its most advanced video intelligence foundation model, on December 1, 2025, at AWS re:Invent.
  • The model is available via Amazon Bedrock and directly through TwelveLabs, with AWS serving as the initial exclusive cloud provider partner.
  • Marengo 3.0 natively processes audio, visuals, movement, and text, offering temporal and spatial reasoning along with new features like sports intelligence for tracking players, teams, and actions.
  • The release offers significant performance gains, including a 50% reduction in storage requirements, twice the indexing speed, up to four hours of video processing per file (double the previous version), and support for 36 languages.
  • With over 50,000 developers on its platform, TwelveLabs aims to help enterprises unlock value from video data, which comprises 90% of digitized content but remains underutilized.

Impact

Marengo 3.0 signals a milestone in enterprise video AI adoption, bringing advanced video understanding and operational efficiencies to mainstream organizations. The exclusive AWS Bedrock partnership is likely to accelerate industry uptake, especially in fields such as media, security, and sports with vast video archives. As enterprises seek ROI from vast stores of video, TwelveLabs is positioning itself at the forefront of next-generation video intelligence solutions.

Read more