Faiss

Facebook’s MIT ANN library for billion-scale embedding search on CPU or GPU.

Install Repository Compare

Fit.

Catalog snapshot

Platforms: Linux, Macos, Windows
Operating systems: Linux, macOS, Windows
Ram floor: 2 GB
Gpu: Optional · 4 GB VRAM helps
Docker: No
Setup: Advanced
Deploy as: Local, Self-hosted
Commercial use: Allowed

Library, not a database: you own persistence and serving.

Project.

Fetched Sep 12, 2026¹

License: MIT
Stars: 40.9k¹
Forks: 4.5k¹
Open issues: 228¹
Language: C++¹
Last commit: Sep 12, 2026¹
Tags: annresearchgpu

About.

Faiss is Facebook’s MIT library for dense-vector similarity search, from million-scale on a laptop to billion-scale on a GPU cluster. It is a C++/Python library with no server, no persistence, and no query language. Raw index power.

CPU and GPU indexes · billion-scale · Python and C++ · IVF, HNSW, and PQ variants

Alternatives.

Nearby.

All Vector databases

TtxtaiAll-in-one Python framework for embeddings, vector search, and RAG workflows.Compare
QQdrantRust vector database with filters, payloads, and a REST API.Compare
WWeaviateVector database with hybrid search and modular vectorizers.Compare
MMilvusCloud-native vector database built for large-scale ANN search.Compare
CChromaDeveloper-friendly embedding database for local RAG prototypes.Compare