Amazon DynamoDB now supports real-time vector search at any scale

This article announces the general availability of vector search in Amazon DynamoDB, allowing users to store vector embeddings alongside operational data and run similarity searches without replicating to a separate vector store.

• Native vector search with single-digit millisecond latency at 99%+ recall, scaling to trillions of vectors
• No servers to provision, patch, or manage; fully serverless with zero downtime maintenance
• Create vector indexes on attributes storing embeddings generated from models like Amazon Bedrock Titan or OpenAI
• Supports up to 4096 dimensions with Euclidean, Cosine, and Dot product distance functions
• Inline filtering enables narrowing search results by non-vector attributes at query time
• SearchVectors API returns results ranked by similarity with optional filter conditions
• Available in all commercial AWS Regions and AWS GovCloud (US) Regions

Vector search in DynamoDB eliminates the need for separate vector databases and synchronization pipelines, reducing operational overhead and costs while maintaining predictable low latency at scale.