turbovec beats FAISS IndexPQ on recall with a data-oblivious rotation, 31GB → 4GB at 10M vectors.

RyanCodrai/turbovec (MIT, Rust core with Python bindings) hit 14,151 stars implementing TurboQuant: a random rotation making every coordinate follow a known distribution regardless of input data, then per-coordinate calibration, Lloyd-Max scalar quantization and bit-packing for 16x compression at 1536 dimensions. Results: 0.2-1.9 points better R@1 than FAISS IndexPQ across 2-bit and 4-bit configs on OpenAI embeddings, 10-19% higher QPS than FAISS FastScan on ARM, ~5% faster on x86 4-bit but slower on x86 2-bit. Integrations for LangChain, LlamaIndex, Haystack and Agno. The data-oblivious property is the real win: it removes the calibration step that makes PQ indexes brittle when your embedding distribution shifts.