Pegasus 1.6 by TwelveLabs

Pegasus 1.6 is a video understanding model by TwelveLabs designed for robotics and physical AI teams. It analyzes egocentric and teleoperated footage from a first-person point of view, recognizing entities and producing persistent metadata across segment types. Teams can feed raw teleoperation or wearable video into the model and receive labeled, timestamped data ready for robot training, without a manual labeling pass.