Announcing Amazon Nova Multimodal Embeddings

AWS has announced Amazon Nova Multimodal Embeddings, a groundbreaking embedding model for semantic search and retrieval across multiple content types.

• First unified embedding model supporting text, documents, images, video, and audio
• Enables cross-modal retrieval with high accuracy
• Supports inputs up to 8K tokens and video/audio segments up to 30 seconds
• Offers multiple embedding dimensions to balance accuracy and performance
• Provides both synchronous and asynchronous API options
• Available in US East (N. Virginia) region through Amazon Bedrock

The model helps break down content silos by mapping diverse content types into a unified space, enabling complex cross-modal searches and applications.