Apple says its most powerful on-device AI model requires an iPhone 17 Pro, iPhone Air, iPad with M4 and later, or Mac with M3 and later, all with 12GB+ of RAM
Apple’s device-side AI eligibility has already tracked memory tiers: related coverage in 2024 tied Apple Intelligence support to the iPhone 15 Pro’s 8GB of RAM while excluding the 6GB iPhone 15. Apple’s newer M-series Macs and iPads have also been introduced with substantially higher memory ceilings and dedicated Neural Engines.
The newly stated 12GB minimum makes that hardware dependency more explicit for Apple’s highest-end local model, extending the gap beyond an initial iPhone generation distinction to a defined set of newer Pro, Air, iPad, and Mac configurations.
First-order effects
Owners of otherwise recent Apple devices below the listed chip and 12GB-memory thresholds cannot run Apple’s most powerful on-device model; eligible-device owners receive a differentiated AI capability.
Apple ties the flagship local-AI experience directly to specific recent hardware tiers, especially higher-memory iPhones, iPads, and Macs.
Second-order effects
The requirement creates a clearer upgrade incentive for users whose devices remain functional but fall short of the memory threshold, reinforcing the value of Apple’s higher-spec product configurations.
Developers building around Apple’s strongest local model will need to treat access as segmented by device eligibility, rather than assume a uniform Apple Intelligence baseline across supported platforms.
Third-order effects
If Apple continues raising local-model requirements, RAM capacity and neural-processing hardware may become a more durable dividing line in device capability and replacement cycles, not merely a specification buyers overlook.
The pattern points to a split AI product strategy: broad features can serve a wider installed base, while the most capable private/on-device workloads remain concentrated on newer hardware.
The trend: On-device AI is turning memory and specialized silicon into explicit feature gates, making hardware configuration increasingly central to software differentiation.
Last time around Apple released a lot of information about how their AI version of Siri worked between local and cloud models, not so much this time It is nice to have a Gemma-like model on device, but it is extremely limited unless it can call a smarter cloud model when needed.
GGs you need 12GB RAM and an M3/A19 Pro or higher to run the most powerful Apple intelligence on-device model. This is mainly used for Siri's advanced voice, but I believe it could also help speed up image understanding instead of relying on Private Cloud Compute.
Apple Intelligence features are available on iPhone 15 Pro and later, but the most powerful on-device model is limited to newer iPhones, iPads, and Macs. Developers are getting access to Siri AI today, but it is not going to be available in the EU initially. [image]
🚨 do you understand what happened to Apple.. At WWDC, Apple confirmed that Siri's cloud layer now runs on a custom Google Gemini model with 1.2 trillion parameters, licensed for roughly $1 billion a year. Strategically, this is a reversal. The company that built its brand on [vid…
What features mandate the “most powerful on-device model” vs. what can be shipped off to the cloud will be interesting to see. In any case, it's all about RAM. Efficiency let Apple cheap out and boost margins for years but there's no way around the need for more where LLMs are …
AFM Core Advanced on-device model running on A19 Pro is a sparse model. It's 20B parameters. It's fully Apple designed. It is an MoE but when it processes the prompt, it only loads the parameters needed and locks them in. If it's 20B parameters total, but on a specific
Meaning, given the fact that shelf life is one of the things people love about Apple, Siri will remain crap for millions. Guys, do you know how badly you are already losing on your own devices. [embedded post]