Faiss

Faiss (Facebook AI Similarity Search) is Meta's library for efficient similarity search and clustering of dense vectors. Handles vector sets of any size, including those exceeding RAM capacity. Faiss 1.10.0 (2025) integrates NVIDIA cuVS library enabling 12× faster index builds and 8× lower search latencies at 95% recall on GPU. Includes GpuIndexCagra, a state-of-the-art graph-based GPU index. CAGRA graphs can be formatted as HNSW for CPU deployment. MIT licensed. Common applications: recommender systems, data mining, text retrieval, content moderation.