All AI news

ElastiCache Serverless adds vector and hybrid search on Valkey 9

AWS announced search and aggregation support on November 10 for AI memory and retrieval workloads, with usage-based charges.

AWS announced vector, full-text and hybrid search plus aggregations for ElastiCache Serverless for Valkey on November 10, 2026. Valkey 9 is required; caches can be created or upgraded through the web console, SDK or CLI. No separate effective date was stated.

Retrieval can support AI agent memory and retrieval-augmented generation (RAG). This is backend infrastructure, not a new chat app or a guarantee of correct AI answers.

Serverless charges cover storage and request processing units (ECPUs), with a 100 MB minimum metered storage amount per Valkey cache. Cost depends on Region and workload; some data transfers may also incur charges. The announcement does not establish Chinese tokenization or Cantonese retrieval quality. No hands-on testing was performed.

References:
https://docs.aws.amazon.com/AmazonElastiCache/latest/dg/search.html
https://aws.amazon.com/elasticache/pricing/
https://docs.aws.amazon.com/general/latest/gr/elasticache-service.html

Usage conditions

Hong Kong
Hong Kong: AWS lists a Hong Kong endpoint and announces search in all commercial Regions. Verify account permissions, Valkey 9 and quotas; this does not confirm activation for every local account.
Macau
Macau: no Macau Region appears in the endpoint table. Confirm account eligibility, Valkey 9, quotas and data-residency requirements for a supported Region. No local-account testing was performed.
Taiwan
Taiwan: AWS lists a Taipei endpoint and announces search in all commercial Regions. Verify account permissions, Valkey 9 and quotas; this does not confirm activation for every local account.

Official source: Amazon ElastiCache Serverless now supports vector, full-text and hybrid searches plus aggregations

Source published:

Source checked:

Published here: