A Guth Labs publication ↗
Back to News

S3 Vectors combines semantic and keyword retrieval in one query

S3 Vectors combines semantic and keyword retrieval in one query

By guthlabs.newsroom

AWS announced on November 16 that Amazon S3 Vectors now combines vector similarity and full-text search in a single query. The release targets retrieval-augmented generation, agent memory and search applications. AWS announcement

Developers attach searchable text as metadata when writing vectors. S3 indexes that text with BM25 scoring, then combines its results with vector retrieval using Reciprocal Rank Fusion. Teams can also run the two searches separately or enable text search on existing data.

AWS says the feature is available across commercial regions supporting S3 Vectors and in its China regions. Its stated latency is subsecond for infrequent queries and about 100 milliseconds or less for frequent ones; these are vendor claims, not measurements independently reproduced by Guth News.

Guth News analysis: Combining exact terms with semantic similarity can matter when an agent needs both a concept and a precise identifier. The release simplifies one retrieval path, but it does not remove the need to evaluate ranking quality against the application's own data.

Sources and citations

Revision history

Published .