01Entry
Laya MLX Runtime
A native MLX runtime for Laya typed decision models running on Apple Silicon, achieving 7-14 ms short decisions without text generation, PyTorch, or cloud APIs. It provides local inference with open weights and benchmark scripts.