New AI startup could shrink server-sized models for use on iPhones
AI-summarised brief · reviewed before publication
PrismML, an AI startup, has developed a method to shrink large language models without affecting performance, reducing a 54GB model to 4GB. Apple, which aims to run advanced AI on devices rather than servers, is exploring a partnership with the company. The technology could enable more powerful AI features on iPhones, though no collaboration is guaranteed.
💡 Why It Matters
- · If successful, this partnership could redefine mobile AI by allowing smartphones to handle complex tasks typically reserved for servers.
- · PrismML’s approach addresses a key limitation—storage and performance—that has hindered on-device AI from reaching its full potential.