Developer uses iPhone as second GPU to speed up MacBook LLM inference

A developer running Qwen 3.8 27B on a 24GB M4 Pro MacBook used an iPhone 17 Pro Max over a 10Gb/s USB-C link as a second GPU to offload some layers, raising end-to-end prefill speed by 29%–44% and moving KV cache beyond 64k context to the phone.

• Eagle-4 Released: Faster than DFlash2 & DSpark2026-10-21
• AnyLanguageModel 1.0 ships for local models on Apple platforms2026-10-20
• Apple unveils new human-crafted Apple Music brand2026-10-21
• PyTorch 2.15 brings FP8 and faster decoding to Mac MPS2026-10-20
• llama server default port changes from 8080 to 99312026-10-20