Apple Releases Free On-Device Model, Won't Rival OpenAI
Apple's free on-device AI model for developers—no token costs, privacy-focused. A game-changer for apps & users.
"AI Disruption" Publication 6800 Subscriptions 20% Discount Offer Link.
On June 10, 2025, the WWDC25 keynote (pre-recorded) has just concluded, and the AI-related segments still occasionally elicit reactions like: “That’s it?”
Many of the so-called new AI features announced by Apple this year, such as real-time translation and Shortcuts, lack revolutionary impact. As for visual intelligence, not only does it lag behind Google Lens by six or seven years in functionality, but its interactive experience also falls far short of the built-in AI/Agent products from various Android competitors in the first half of 2025.
However, having watched the entire event, I can confidently say one thing: there’s one piece of news about Apple Intelligence that will have a profound impact, for both regular users and developers.
That is the “Foundation Models Framework,” Apple’s on-device AI model integration framework for developers.
Using it, Apple developers can easily integrate large language models (LLMs) into their apps on-device, without incurring token costs.
This means developers can effortlessly provide LLM-based capabilities and services to users, performing purely on-device inference without worrying about the token fees associated with calling cloud-based LLM APIs, naturally sparing users from bearing those costs.
Developers can more confidently maintain their preferred monetization models, and users benefit as a result.
