Turbovec – Google's TurboQuant for vector search in Rust

	Turbovec – Google's TurboQuant for vector search in Rust(github.com)
	290 points by fittingopposite 3 days ago \| 34 comments
	tl;dr: Turbovec is a Rust vector index (with Python bindings) implementing Google Research's TurboQuant quantizer, compressing a 10M-doc corpus from 31GB to 4GB while outperforming FAISS IndexPQFastScan by ~3.4× at 4-bit and ~23% at 2-bit across ARM and x86. It supports online ingest with no training phase, SIMD-accelerated search with allowlist filtering, and crash-safe incremental saves. Drop-in integrations exist for LangChain, LlamaIndex, Haystack, and Agno, targeting local/air-gapped RAG stacks.
	HN Discussion: ↓FAISS benchmark comparison is outdated since FAISS is no longer state-of-the-art ↑Excitement about compression ratio and potential for local/lightweight deployments ↓TurboQuant itself has unresolved academic misconduct allegations, tainting this project ↓README appears AI-generated, hurting adoption prospects ↓Alternative tools like Qdrant already integrate TurboQuant, questioning need for this