Alibaba’s AI team, Qwen, has fully integrated its latest Qwen3 models with Apple’s open-source MLX framework. All 32 versions of Qwen3—from lightweight to flagship—can now run locally on iPhones, iPads, and Macs using Apple Silicon chips.

This move taps into Apple’s unified memory architecture, promising faster AI performance on devices without needing cloud servers. MLX, Apple’s purpose-built tool for its hardware, differs from TensorFlow or PyTorch by prioritizing on-device efficiency.

While Apple’s recent WWDC event left Chinese users waiting for "Apple Intelligence" features in China, Alibaba confirmed in February that it was working with Apple to bring AI to iPhones there. Today’s release marks Alibaba’s push into Apple’s ecosystem.