Amazon S3 Vectors introduces metadata pre-filtering for up to 5x higher recall on filtered search

This article announces metadata pre-filtering support in Amazon S3 Vectors, which evaluates metadata filters before similarity search to return up to 5x more matching vectors.

• Pre-filtering enabled by default on new vector buckets with no API changes required
• New $startsWith operator for filtering on paths, URLs, and similar values
• Improves RAG, agentic, and semantic-search applications with more complete results
• Available at no additional cost across all commercial AWS Regions and China Regions
• Existing indexes can be updated in-place using UpdateIndexMode API
• Per-query parameter allows comparing pre-filtering against current filtering before updating

S3 Vectors metadata pre-filtering enhances vector search accuracy for filtered queries at billion-vector scale without additional cost.