S3 Vectors metadata pre-filtering comes to GovCloud, plus a $startsWith operator!
Hi, it's Shii-chan! The S3 Vectors filtering feature has reached GovCloud.
AWS What's NewWhat was announced?
AWS What's New announced that Amazon S3 Vectors metadata pre-filtering is now available in the AWS GovCloud (US-East) and AWS GovCloud (US-West) Regions.
What changes
Pre-filtering evaluates metadata filters before running similarity search. When your filter is selective, it returns up to 5x more of the matching vectors. RAG, agentic, and semantic-search applications get more complete results when they filter, so they can return more relevant answers.
Dive Deep
Details:
- A new prefix match operator, $startsWith, lets you filter on values like paths and URLs
- Indexes in new vector buckets in the GovCloud (US) Regions use pre-filtering by default, with no change to how you write vectors with PutVectors or run filtered queries with QueryVectors
- To use pre-filtering on an existing index, update it in place with the UpdateIndexMode API
- Before updating, you can compare pre-filtering against your current filtering on the same index using a per-query parameter on QueryVectors
- It's available at no additional cost in all commercial AWS Regions where S3 Vectors is available, and in the AWS China Regions
The change is still being deployed, with completion planned in the coming days. To get started, use the AWS CLI, AWS SDKs, or the Amazon S3 console.
Wrap-up
- Pre-filtering is available in GovCloud (US-East / US-West)
- Up to 5x more matching vectors when a filter is selective
- New $startsWith prefix match operator
- Update existing indexes with UpdateIndexMode, and compare beforehand
It's a good fit if you run RAG or semantic search in GovCloud.