More-efficient approximate nearest-neighbor search
20% to 60% faster. Amazon's new graph search method is rewriting the rules of AI vector databases.

Why it matters
This breakthrough in approximate nearest-neighbor search could significantly reduce compute costs for AI applications that rely on vector similarity search, from recommendation engines to RAG systems.
The key facts
4 to know20% to 60% speed improvement
Graph-based search optimization
Works regardless of graph construction method
Amazon Science research
Go to the source
Amazon Scienceamazon.science
Publisher excerpt: New approach speeds graph-based search by 20% to 60%, regardless of graph construction method.


