Amazon S3 Vectors now supports up to 40 KB of filterable metadata per vector, up from 2 KB. You can attach and filter on more attributes per vector, such as category, permissions, timestamps, and descriptive tags, so your retrieval-augmented generation (RAG), agentic, and semantic-search applications can narrow results by more of the fields that matter to them.
S3 Vectors provides native support to store and query vectors in Amazon S3, delivering purpose-built, cost-optimized vector storage and query at billion-vector scale. With this launch, indexes in new vector buckets support the higher limit by default, with no change to how you write vectors with PutVectors or run filtered queries with QueryVectors. To use the higher limit on an existing index, update it in place to enhanced index mode with the UpdateIndexMode API.
The higher filterable metadata limit is available at no additional cost in all AWS Regions where Amazon S3 Vectors is available. To get started, use the AWS CLI, AWS SDKs, or the Amazon S3 console. To learn more, visit the Amazon S3 Vectors documentation.
Free. Unsubscribe any time.
Related updates
Amazon S3 Vectors introduces metadata pre-filtering for up to 5x higher recall on filtered search
Amazon S3 Vectors now supports up to 10,000 similarity search results per query
Amazon S3 Vectors is now generally available with 40 times the scale of preview
Generally Available: Filtered vector search in Azure Cosmos DB for MongoDB (vCore) · Azure
AI Search - Store larger custom metadata values in AI Search · Cloudflare