We understand that on-device Foundation Models (FMs) can evolve between OS releases. To help us accurately scope our application capabilities and performance expectations for multiplatform development, could you clarify the variation of these new on-device models across different hardware? Specifically:
- Within the same OS & device family: Do the architecture, parameters, or capabilities of the on-device models vary based on hardware tiers (e.g., iPhone vs. iPhone Pro, or MacBook Air M5 vs. MacBook Pro with M5 Pro)?
- Across different device form factors: Are there model variations between hardware families running equivalent OS releases (e.g., Mac vs. iPhone)?
Knowing if we are targeting a uniform model baseline or a tiered model ecosystem will greatly help us optimise our App Intelligence features or at least set us with proper expectations in scope and capabilities.
Thanks.
Within the same OS & device family: Do the architecture, parameters, or capabilities of the on-device models vary based on hardware tiers (e.g., iPhone vs. iPhone Pro, or MacBook Air M5 vs. MacBook Pro with M5 Pro)?
Yes. This post from Apple's Machine Learning blog describes our latest on-device models. There is AFM 3 Core and AFM 3 Core Advanced.
Up until this point we've had the same on-device model across all devices, with different model versions mentioned here in the docs for SystemLanguageModel.
Starting in the fall with the release of Siri AI, you can read in this press release:
Apple’s most powerful on-device model and the features it enables, like expressive voices and more advanced dictation, are available on iPhone Air, iPhone 17 Pro, iPhone 17 Pro Max, iPad (M4) or later with at least 12GB of unified memory, and Mac (M3) or later with at least 12GB of unified memory. The model and expressive voices are also available on Apple Vision Pro (M5).
Essentially the most powerful devices will be able to run AFM 3 Core Advanced while all other devices will just be able to run AFM 3 Core.
So in short, yes! Plan to have different models. But expect model details and guidance to evolve over this summer's beta period.
