Developer uses an iPhone 17 Pro Max as a second GPU for a MacBook Pro

On October 3, 2026, cnBeta reported on an experiment by a developer who used custom software to connect an iPhone 17 Pro Max to a MacBook Pro powered by the M4 Pro chip, letting the phone act as a second GPU for running local large language models. According to the report, running Qwen3.8-27B on this setup improved prefill performance by up to 44%.

The report presents the work as an individual developer's experiment rather than a product launch or an officially supported feature from any vendor. Beyond the fact that the link is handled by custom software, that the phone takes on the role of a second GPU, and that the measured gain on Qwen3.8-27B reaches 44% in prefill, the report offers no name for the custom software, no details on the physical or software interface between the two devices, and no information about the context length or batch size used in testing. It also does not say whether the code has been released, whether the result is reproducible, or who the developer is.

As it stands, the story remains at the stage described in that single report: a developer-run experiment with no vendor response and no reported follow-up.