infinity

Why we included this project

Infinity is a database built around hybrid search: it runs dense and sparse vector, tensor, and full-text queries in one engine, and lets you blend the results with built-in rerankers like RRF or ColBERT instead of writing that logic yourself. That combination matters for retrieval-heavy LLM apps such as RAG, copilots, and question answering, where you often want both semantic similarity and exact keyword matching on the same data. The single-binary design with no external dependencies and a Python API keeps deployment simple, whether you embed it in your app or run it as a server. If you're building a proof of concept for hybrid retrieval and want to compare it against a hand-rolled stack, it's a convenient starting point.