Alibaba has released Qwen3 AI models optimized for Apple's MLX architecture, enabling deployment across iPhones, iPads, MacBooks, and Macs. The series includes six dense models (ranging from 0.6B to 32B parameters) and two Mixture-of-Experts models (30B and 235B). Qwen3 features hybrid reasoning that switches between intensive "thinking" mode for complex tasks and faster "non-thinking" mode for simple responses. The models are designed for versatility, supporting deployment on consumer devices, smart glasses, autonomous vehicles, and robotics. By making these open-source models compatible with Apple's architecture, Alibaba aims to broaden AI adoption across Apple's ecosystem and empower global developers to create advanced AI applications.