Search in Python

The elasticsearch and opensearch-py packages are the official low-level clients for their engines. OpenSearch began as a fork of Elasticsearch, and the two APIs have drifted apart over time, so use the client that matches your cluster. Meilisearch is a simpler engine with typo tolerance and good defaults, and its official client makes it quick to add search to a site or app. django-haystack gives Django a single search API with pluggable backends, which keeps your views independent of the engine underneath.

Faiss, from Meta, is a library for similarity search over dense vectors. It runs in memory and offers many index types, from exact flat search to approximate methods that trade a little accuracy for large speed and memory gains. The faiss-cpu package is its CPU build. If you need a persistent vector database rather than an index library, see the Database section.