Yhn TechnologyMobile first seen 3 d ago, last 35 min ago, peak #8
Qwen model reportedly runs on iPhone Neural Engine at 15-70 tokens per second
Original: Qwen3.8-27B on the iPhone 18 Neural Engine at 15–70 tok/s (1.5 bpw)
A demonstration claims a 27-billion-parameter Qwen large language model is running on an Apple iPhone's Neural Engine, generating text at 15-70 tokens per second using extreme quantization (1.5 bits per weight). If accurate, it would put a large AI model entirely on a phone without cloud access. The claim has drawn attention from people tracking on-device AI, though the hardware details could not be independently verified.
Why now: unclear
Rank over time, top of the chart is #1. 48 snapshots from 3 d ago to 35 min ago.
Evidence
API: https://socialmediatrends-api.osmike.com/v1/trends/2642601