Amazon Bedrock Managed Knowledge Base now supports multimodal embeddings for video, audio, and image content with TwelveLabs Marengo 3.0

AWS announces TwelveLabs Marengo 3.0 embedding model integration with Amazon Bedrock Managed Knowledge Base for multimodal content search.

• Create multimodal embeddings for video, audio, and image content without transcription limitations
• Encode visual scenes, speech, and video cues directly into 512-dimensional vectors
• Search media assets using natural language with segment timestamps for precise navigation
• Upload from data sources like Amazon S3 with no infrastructure management required
• Enables use cases in sports analytics, media, security, education, and retail
• Configurable segmentation options to match content structure

Marengo 3.0 delivers state-of-the-art retrieval accuracy for multimodal media search across diverse industries.