SlotStream

Overview

A Mac-native tool built with Swift and MLX that uses expert offloading and SSD streaming to run large Qwen models on limited RAM. Devs should care because it makes local large-model inference practical on consumer hardware without relying on cloud GPUs.