Apple's soon-to-be-released macOS Tahoe 26.2 is set to bring about remarkable improvements in machine learning performance for Macs in the M5 series. This pivotal upgrade is a result of Apple's efforts to fortify its in-house, open-source machine learning framework, MLX. Tailored specifically for Apple Silicon, the upgraded MLX will offer native support for the neural accelerators embedded within each GPU core of the M5 chip. As a consequence, when handling machine learning tasks, the M5 Macs will be capable of achieving response speeds that are up to four times faster compared to those of the M4 models.

Moreover, MLX has rolled out new support for Thunderbolt 5 clusters. This innovation enables multiple Macs to share memory resources and engage in distributed computing. Nevertheless, it's worth noting that currently available M5 models only support Thunderbolt 4. Therefore, the full potential of cluster acceleration won't be unlocked until the M5 Pro and M5 Max models, which come equipped with Thunderbolt 5, are launched in early 2026.