Apple evaluates PrismML technology to run 27B parameter AI models on iPhone – 4sysops
The primary objective of this collaboration is to shift complex AI processing from the cloud to local device hardware to enhance user privacy and reduce latency. By executing large models natively, Apple can. . .
The primary objective of this collaboration is to shift complex AI processing from the cloud to local device hardware to enhance user privacy and reduce latency. By executing large models natively, Apple can provide advanced reasoning and software development assistance without an active internet connection or expensive server infrastructure. PrismML claims its compressed models are six to eight times faster and significantly more energy-efficient than standard versions, though they may experience a slight decrease in factual recall. Apple evaluates PrismML technology to run 27B parameter AI models on iPhone (image cnbc. com)While discussions are in early stages, the technology could fundamentally change how system resources are allocated for on-device intelligence. Beyond smartphones, these compression techniques have implications for any edge computing environment, including robotics and autonomous systems, where local decision-making is critical. Industry analysts note that while this reduces memory requirements per task, it may shift chip demand from datacenters to consumer devices rather than decreasing overall semiconductor reliance. Apple in talks with startup that shrinks AI models to run on an iPhone | cnbc. comApple in Talks to Acquire PrismML, Startup That Shrinks AI Models to Run on iPhone | mlq. aiApple eyes PrismML to run huge AI models directly on … | macdailynews. com