You can now run Max AI models on Apple Silicon (forum.modular.com)

Apple's MAX AI models can now run on M1 to M5 Apple Silicon GPUs, marking a significant advancement for developers as they can leverage Apple's hardware for machine learning tasks. The latest 26.4 release enhances support for various models, including text language models, vision models, and image diffusion, allowing them to execute efficiently on these devices, particularly benefiting from the dedicated matrix-multiplication operations in M5's Neural Accelerators. This breakthrough is crucial for the AI/ML community as it opens up new possibilities for running complex models directly on Apple hardware, without requiring external cloud resources. Users can easily deploy models using simple command-line instructions, making it more accessible for developers to test and utilize AI capabilities. While older M-series chips may encounter some compatibility issues due to limited testing, the ongoing development indicates a commitment to optimizing performance across all Apple Silicon systems. The announcement highlights the potential for improved benchmarks as developers fine-tune operations, making it an exciting time for AI experimentation on consumer devices.