Faiss

Faiss is Meta's Fundamental AI Research library for efficient similarity search over dense vectors — the engine inside a large share of the world's vector search deployments. It implements exact search plus approximate indexes (IVF, HNSW, product quantization, binary codes) that trade recall for speed and memory, scaling to billions of vectors on a single server, with GPU implementations that drop in as replacements for CPU indexes. Written in C++ with full Python/numpy wrappers; CUDA and AMD ROCm support. It is a library, not a database: no persistence or replication — pair it with your own storage or a vector DB for production.

```bash

# via conda (primary channel) or pip

pip install faiss-cpu

```

```python

import faiss, numpy as np

index = faiss.IndexFlatL2(128)

index.add(np.random.random((10000, 128)).astype("float32"))

distances, ids = index.search(np.random.random((5, 128)).astype("float32"), k=10)

```

GPU builds: faiss-gpu / faiss-gpu-cuvs on Anaconda. Wiki tutorial: github.com/facebookresearch/faiss/wiki.