Send a message to run inference on a provider's Mac via MLX. Responses are powered by decentralized compute.