Amazon Bedrock Knowledge Base adds TwelveLabs Marengo 3.0 multimodal embeddings

AWS said Amazon Bedrock Managed Knowledge Base now supports TwelveLabs Marengo 3.0 as an embedding model, letting customers encode video, audio and image content directly rather than relying only on transcription. The company said the model produces compact 512-dimensional vectors and returns segment start and end times so applications can jump to a relevant moment in a video. Media can be uploaded from sources such as Amazon S3, synced and searched in natural language with no infrastructure to manage, AWS said. AWS listed sports analytics, media and entertainment, security, education and retail as intended uses, and said segmentation options are configurable.