TwelveLabs Releases Pegasus 1.6 AI Model for Enhanced Physical World Understanding
TwelveLabs has launched its Pegasus 1.6 model, designed to provide machines with a deeper understanding of real-world environments through video analysis. This new version focuses on egocentric video, enabling robots, drones, and autonomous vehicles to perceive, reason, and act more effectively. Pegasus 1.6 offers features such as action segmentation, dense captioning, quality scoring, search and curation, and consent flagging, aiming to improve robot training data and operational safety by interpreting human actions and interactions from a first-person perspective.
Why it matters
Pegasus 1.6 enables machines to perceive, reason, and act in the physical world by understanding egocentric video.
Source: The Robot Report — read the original
Source published 6 Oct 2026, 20:07 UTC