Apple Caps iOS 27 AI with Timeouts and Extra Costs

The new Apple Intelligence features in iOS 27 come with safeguards: usage caps, timeout periods, and the option to pay for extended access. Apple states that these limits, based on server-side models, vary depending on the feature, the complexity of requests, and system load. An initial scale is already visible in the Home app, indexed on iCloud+.
iCloud+ sets the number of AI cameras in Home: 1, 2, or 5
In the Home app used with security cameras, the number of cameras that can benefit from Apple Intelligence features depends on the iCloud+ plan. With the 2 TB plan, AI applies to one camera. The 6 TB plan allows for AI on two cameras. The 12 TB plan raises this cap to five cameras.
Usage caps, timeouts, and paid access for overages
Apple will limit the use of Siri AI and other Apple Intelligence features. When a limit is reached, a timeout period is triggered, and the affected functionality becomes temporarily inaccessible. During this period, it remains possible to pay to continue using the feature beyond the allotted quota.
Le brief IA que les pros lisent chaque soir
Les 7 actus IA du jour, décryptées en 5 min. Gratuit.
Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.
Choisissez votre rythme
Gratuit · Pas de spam · Désabonnement en 1 clic
Affected services include Siri AI, Photos, Image Playground, and Shortcuts
The intelligent photo editing tools Clean Up, Extend, and Spatial Reframing are subject to usage limits, as is Image Playground. The AFM 3 Cloud and AFM 3 Cloud Pro models used in Shortcuts, as well as developer applications relying on Apple Foundation Models in Private Cloud Compute, are also affected. Siri AI corresponds to the new version of the voice assistant, while the photo tools allow for element removal, framing or perspective adjustments, and Image Playground is used to generate or transform images.
Server-side models and variable rules based on load
Apple clarifies that the Apple Intelligence features in iOS 27 rely on powerful models executed server-side, with limits established to ensure a reliable experience. These caps may vary depending on the feature, the complexity of requests, real-time server load, system policies, and other factors. Generative AI is primarily processed in the cloud, and high demand can degrade performance. The AFM 3 Cloud and AFM 3 Cloud Pro models operate on servers: the former, dubbed the "workhorse," aims for speed and efficiency; the latter, more robust, is intended for demanding tasks such as agentic tools and complex reasoning.
Brief IA — L'actualité IA en français
L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.