Turbovec – Google's TurboQuant for vector search in Rust

Turbovec is a Rust-based vector search library implementing Google Research's TurboQuant algorithm, offering significant memory efficiency and speed advantages over existing solutions like FAISS. It can compress a 31 GB dataset to 4 GB while searching faster, features online ingestion without training phases, hand-optimized SIMD kernels for multiple CPU architectures, and includes Python bindings for easy integration. The library prioritizes privacy by running entirely locally and supports incremental saves and runtime filtering capabilities.