Infino
An embedded retrieval library that runs full-text, vector, hybrid and SQL search over one table, stored as ordinary Parquet files on local disk or object storage.
Infino is an embedded retrieval library offering full-text search, vector search, hybrid search and SQL over a single table, stored as ordinary Parquet files rather than a proprietary format, on local disk or in object storage. It is distributed as a library for Python, Node.js and Rust rather than as a server process.
Its own published benchmarks report sub-millisecond warm queries for vector and full-text lookups at one million documents on tables held in object storage, rising to single-digit milliseconds at ten million documents, with cold first-query latency dominated by opening files and filling caches. Engine behavior, including vector index settings, is configured only through a YAML file rather than environment variables.
It installs the mimalloc memory allocator globally by default, which can be turned off when it is embedded inside a process that already sets its own allocator.