Apple is negotiating with the Silicon Valley startup PrismML to implement powerful artificial intelligence models in the iPhone. This was reported by CNBC.
PrismML's CEO Babak Hasibi confirmed that the tech giant is already testing the solutions developed by the company.
"They are currently evaluating our technology," he noted, adding that the negotiations are still in the early stages.
PrismML, which raised $16.25 million in March, has managed to reduce the size of Alibaba's open Qwen model from 54 GB to less than 4 GB. This enables the iPhone 15 to run all 27 billion parameters of the model.
Experts estimate that local data processing will reduce the costs of cloud computing and increase user data privacy levels.
Meanwhile, Morgan Stanley predicts that the average cost of operational memory for Apple could increase by about 190% in the 2027 financial year, which could lead to a price increase of around $200 for the base models of the iPhone 18.
The compression technology developed by PrismML is based on simplifying the storage of internal model data. Developers claim that such models use 10-15 times less memory, generate responses 6-8 times faster, and consume 3-6 times less energy.
This technological advancement occurs against the backdrop of the release of the public beta version of iOS 27, in which Apple is trying to make the Siri voice assistant more competitive compared to OpenAI's solutions.
Technological publications are supported by Idram.
