A New Era for Apple Software

During the recent Worldwide Developers Conference (WWDC), Apple unveiled its highly anticipated artificial intelligence strategy. The highlight of the event was the introduction of 'Apple Intelligence' and the reveal of five new AI foundational models, developed in a strategic partnership with Google. These innovations are set to power a revamped version of Siri, alongside enhancements for MacOS, Vision Pro-integrated tools, and upcoming iOS features.

«We believe that truly helpful AI must be centered around you and your needs,» stated Craig Federighi, Apple’s senior vice president of software engineering, during the keynote presentation.


The Five New Foundation Models

Apple has introduced a suite of models designed to handle various tasks, ranging from local on-device processing to complex cloud computing:

  • AFM 3 Core: A 3-billion parameter base model built specifically for local execution on iPhone and Mac devices.
  • AFM 3 Core Advanced: A 20-billion parameter multimodal model, positioned as Apple’s most capable on-device solution.
  • AFM 3 Cloud: A version optimized for server-side cloud operations.
  • AFM 3 Cloud (Image): Specialized in image processing to support AI photo editing tools and the 'Image Playground' suite.
  • AFM 3 Cloud Pro: The premium cloud model designed for complex reasoning and demanding agentic tasks.

The Role of Siri and Privacy Standards

While these models serve as the technological foundation—the 'roads'—the user-facing experience is defined by the new Siri. This upgraded assistant will be deeply integrated across the Apple ecosystem, including iPhones, Macs, Watches, and AirPods. A core promise from Apple is data privacy; the company asserts that it does not store user data or interaction logs, maintaining these standards even for cloud-based tasks processed via Google Cloud infrastructure and Nvidia GPUs.


Privacy and Data Sources

Apple clarified that its models are trained using a combination of publicly available data, licensed content, and synthetic data. The company explicitly stated that it does not utilize private user data or interactions for model training and claims to respect the preferences of web publishers who choose to opt out of such practices.


Strategic Implementation

Unlike some competitors, Apple is emphasizing on-device AI to ensure speed and privacy. While some features, like an improved Siri, appear to be evolutionary rather than revolutionary, the integration of multimodal capabilities—where the AI can process both text and images—marks a significant technical leap. According to Francisco Jeronimo, vice president of client devices at IDC, Apple’s strategy is to make AI 'invisible' and deeply woven into the operating system.

«The impact could be significant. If Apple makes AI feel natural, private and useful for mainstream users, it will not just strengthen its ecosystem. It could redefine what consumers expect from every device they use,» Jeronimo noted.