FAISS

FAISS (Meta) is a high-performance similarity search library (not a full server DB) for dense vectors, often embedded in apps or wrapped by services. Algorithmic core under many vector stacks. Use for raw in-process speed or as a library dependency beneath higher-level stores. Operators evaluating FAISS should check embedding/index APIs, persistence and filter semantics, and how retrieval latency shows up under the load that matters for their corridor—not only demo latency.