TwelveLabs Pegasus 1.5 video-to-text model lands on Amazon Bedrock

Pegasus 1.5 is now available on Bedrock as a managed model for whole-video understanding.

Instead of analyzing individual frames, it processes the full video and produces a time-coded record against a schema the user defines.

The release complements AWS's Elemental AI tools announced at IBC and competes with Gemini's native video understanding.

The open questions are cost per hour of video and how accurate the timestamps are.