Vector compression with TurboQuant codecs for embeddings, retrieval, and KV-cache. 10x compression, pure NumPy core — optional GPU acceleration via PyTorch (CUDA/MPS) or MLX (Metal).

Compress KV cache for LLM inference with 5.02x efficiency on NVIDIA GPUs using cuTile kernels.

Compress embeddings, retrieval vectors, and KV-cache with TurboQuant codecs for 10x smaller storage and NumPy-first AI workloads

Compress MLX KV cache on Apple Silicon with TurboQuant mixed-precision and fused Metal kernels for lower memory use and fast decode