Fits 10 million documents into 4GB of RAM

turbovec just dropped: a Rust vector index built on Google TurboQuant (ICLR 2026). Zero codebook training, 16x compression, beats FAISS on ARM with AVX-512 kernels. 10M docs that cost 31 GB as…

A 10M document corpus eats 31 GB of RAM as float32 Most teams hit that wall & reach for a managed vector database. $400 a month or $800 a month. Scaling with every document you add. Then a solo…