Skip to main content
All terms
Frameworks & Tools

FAISS

A library from Meta for fast similarity search over large collections of vectors.

Definition

FAISS is an open-source library from Meta for efficient similarity search and clustering over large sets of dense vectors. It provides indexing methods that find the nearest embeddings to a query, trading exact results for speed and memory when needed, and can run on CPUs or GPUs. It is a foundational building block inside many vector databases and retrieval systems.