facebookresearch/faiss

Health v2 · maintenance only

Adoption not part of the score

40809 stars · 4505 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Faiss is a C++ library (with Python wrappers) for efficient similarity search and clustering of dense vectors, supporting indexes that scale to billions of vectors including GPU-accelerated implementations. It provides approximate and exact nearest-neighbor search with L2, inner product, and cosine similarity metrics.

Use cases

• find nearest neighbors of embeddings
• build semantic search over dense vectors
• similarity search at billion-vector scale
• gpu-accelerated vector search
• cluster high-dimensional vectors
• knn search for recommendation systems
• vector index for RAG retrieval

When to choose

• you need fast approximate or exact nearest-neighbor search over dense embeddings
• your dataset is too large to fit in RAM or needs GPU acceleration
• you want a mature, battle-tested library with Python bindings

When to avoid

• you need a full managed vector database with CRUD, filtering, and replication
• your vectors are sparse or textual rather than dense embeddings
• you need real-time updates and deletions, which Faiss indexes support only in limited ways

Facets

library · maturity stable

vector-database search-engine machine-learning gpu-computing machine-learning databases artificial-intelligence cpp python windows cross-platform nearest-neighbor-search ann similarity-search dense-vectors embeddings hnsw clustering cuda faiss search retrieval-augmented-generation gpu linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="facebookresearch/faiss")