Apple caps iOS 27 AI features and charges for overuse

Apple has published an official support document confirming that the new Apple Intelligence features in iOS 27 will carry usage quotas, and users who exceed them will either wait or pay. The company says the models run in the cloud, so limits are necessary to keep performance stable for everyone. This is the first time Apple has attached a price tag, even an indirect one, to heavier use of the upgraded Siri and its generative tools.

The quota covers Siri AI, the new version of the voice assistant, alongside the smart image-editing tools — Clean Up, Extend and Spatial Reframing — the Image Playground photorealistic image generator, and the AFM 3 Cloud and AFM 3 Cloud Pro models exposed through Shortcuts. Developer apps that call Apple Foundation Models via Private Cloud Compute draw from the same pool. Apple describes AFM 3 Cloud as a "workhorse" optimized for speed and efficiency, while the Pro variant targets more demanding workloads such as agent tooling and complex reasoning.

Apple does not disclose a precise formula. Limits vary by feature type, request complexity, current server load, system policy and other factors. When a user hits the ceiling, the specific feature enters a "timeout" and remains blocked until the limit resets. Users who need immediate access can buy an extension, though pricing and the pricing model have not been published. The approach lets Apple gauge real-time capacity and calibrate policy on the fly.

The document provides a single numerical example: in the Home app, the number of security cameras that can be monitored through Apple Intelligence depends on the iCloud+ tier. A 2 TB plan allows one camera, 6 TB allows two, and 12 TB allows up to five. It is the first hint of how Apple may link cloud storage volume to AI compute allowance, a model that could extend to other services.

The vague wording is deliberate. Apple prefers not to commit to hard thresholds at launch, preserving flexibility once real-world loads become clear. For users, that means uncertainty about exactly when access will be cut off and how much it will cost to restore it. For developers building on Apple's models, the same uncertainty complicates cost planning and end-user pricing.