Amazon S3 Vectors adds hybrid search, combining vector and full-text matching in one query
Pure vector search misses exact identifiers such as product codes, error strings and names. Keyword search misses paraphrases. Hybrid retrieval has become standard practice, but teams usually run two systems and fuse the results themselves. S3 Vectors now handles that fusion natively.
This follows ElastiCache Serverless adding vector plus BM25 hybrid search in the same weekly roundup. AWS is spreading hybrid retrieval across its storage tiers, from cheap and durable S3 Vectors to low-latency ElastiCache.
Competitively, this pressures specialized vector databases such as Pinecone, Weaviate and Qdrant, which have used hybrid search as a differentiator. S3's appeal is cost and simplicity at large scale, while dedicated engines still win on latency and tuning control.
Watch for details on ranking fusion (reciprocal rank fusion or learned weighting), how much control developers get over weights, and the query pricing impact.