Apple ML Research: Muse Spark Open with MLX on Mac Studio

Apple has released a new version of MLX that enables researchers and developers to run large language models locally on Mac Studio hardware, including the trillion-parameter Muse Spark Open model. Using Apple's M5 Ultra chip with GPU Neural Accelerators and unified memory architecture, a single Mac Studio can run these models privately, and multiple Macs can be clustered together via RDMA over Thunderbolt to achieve inference speeds exceeding 36 tokens per second—well above human reading speed.