Qwen 3.5, a two-billion parameter AI model optimized with Apple’s MLX framework, has demonstrated significant advancements in on-device AI by running efficiently on the iPhone 17 Pro. This model, utilizing 6-bit quantization, showcases strong visual intelligence and outperforms larger models, making powerful AI more accessible on mobile devices. The development reflects a broader trend in mobile AI, enabling sophisticated language models to operate directly on smartphones and opening opportunities for innovative new products.
Qwen 3.5 2B outperforms larger models on iPhone 17 Pro
