Apple Evaluates PrismML Tech for On-Device iPhone AI

Apple is in discussions with PrismML, a Silicon Valley startup and Caltech spinout, to evaluate technology that compresses large artificial intelligence models for local execution on iPhones. The collaboration aims to move complex AI tasks, including health tools and video generation, off the cloud and directly onto the device to reduce latency and cloud costs while improving user privacy.

PrismML recently demonstrated its capability by shrinking Alibaba's open-source Qwen model from 54 GB to under 4 GB. This compression allows the model's 27 billion parameters to run on an iPhone 15 or newer. PrismML CEO Babak Hassibi confirmed that Apple and other firms are currently testing the models for performance, energy efficiency, and speed.

These discussions occur following the public beta launch of iOS 27. The move is part of a broader effort by Apple to make Siri more competitive with AI assistants developed by Anthropic and OpenAI by increasing the capabilities of on-device processing.