DynamoDB Introduces Native Vector Search Capabilities

Amazon DynamoDB has introduced native vector search capabilities, allowing users to store and query large-scale datasets using similarity searches.

The new feature supports real-time vector search at any scale, with single-digit millisecond latency and 99%+ recall. Vector indexes have no storage limits and can scale horizontally as data grows.

To use vector search in DynamoDB, users must create a vector index on an attribute storing vector embeddings. The process involves generating embeddings using a model of their choice, such as Amazon Bedrock Titan Text Embeddings or OpenAI text embedding models, and storing them alongside existing operational attributes.

The SearchVectors API accepts a query vector, the number of results to return (up to 100), and optional filter conditions. It returns results ranked by similarity.