Apple tests daily limits on Siri AI and Apple Intelligence in iOS 27.3
Apple has started testing daily usage limits for Siri AI and Apple Intelligence features that run on its servers, in the iOS 27.3, iPadOS 27.3 and macOS 27.3 betas released on November 16, 2026, MacRumors reports. Simple Siri commands stay free and unlimited, but multi-step Siri requests, image tools, some Shortcuts and third-party apps that use Apple’s cloud models get a free daily allowance, and paid iCloud+ plans raise it. For developers who built on Apple’s server model, it is the first concrete look at the quota their users will hit.
What’s new
Per MacRumors, four categories of features count against the free daily limit:
- Siri AI multi-step, personal context requests that need server-side processing, such as asking when you need to leave home to arrive somewhere on time.
- Image generation and editing: Image Playground and the AI tools in Photos, including Clean Up, Extend and Spatial Reframing.
- Shortcuts workflows and automations that use AFM Cloud and AFM Cloud Pro, Apple’s cloud Foundation Models.
- Third-party apps that use Apple Foundation Models in Private Cloud Compute.
Single requests such as “set a timer for 30 minutes” don’t count. Each category has its own limit, and Apple hasn’t said how many requests any of them allow; MacRumors says limits depend on the feature, how complex the request is and system demand, and may change. The free limit is per device and doesn’t roll over. It resets at about midnight in the user’s time zone, and Apple plans to warn users as they approach a limit and notify them when they reach it.
Existing iCloud+ plans get more usage at no change in price:
| iCloud+ plan | Price per month | Daily limit |
|---|---|---|
| 50GB | Not stated | No increase |
| 200GB | $2.99 | Up to 2x |
| 2TB | $9.99 | Up to 5x |
| 6TB | $29.99 | Up to 10x |
| 12TB | $59.99 | Up to 20x |
A device uses its free allowance first, then the plan’s. Family members share one pool, first come, first served, reset in the plan owner’s time zone. Apple isn’t offering usage beyond the 12TB plan for now. The Verge notes that last week Apple added $49.99 Intelligence Usage Packs to Logic Pro’s assistant, with other Creator Studio apps to follow.
What it means for developers
If your app calls Apple’s server model through the Foundation Models framework, your users’ requests now fall under a daily limit: third-party apps on Apple Foundation Models in Private Cloud Compute are one of the four limited categories, with a limit of their own. Apple laid out the model in its WWDC26 session Build with the new Apple Foundation Model on Private Cloud Compute: there are no token costs and no API keys for the developer, each user gets a daily limit, users can upgrade to iCloud+ for more, and the server model is open to apps with fewer than 2 million downloads that apply for access.
The same session shows how to handle the limit instead of surfacing a raw error. PrivateCloudComputeLanguageModel exposes quotaUsage: check isLimitReached, read isApproachingLimit from the .belowLimit status to warn early, and call show() on limitIncreaseSuggestion to open Apple’s upgrade options. Apple’s sample:
struct ArticleSummarizationView: View {
private var model = PrivateCloudComputeLanguageModel()
var body: some View {
if case .belowLimit(let info) = model.quotaUsage.status {
if info.isApproachingLimit {
Text("Nearing usage limit.")
.foregroundStyle(Color.orange)
}
}
if model.quotaUsage.isLimitReached {
Text("Usage limit exceeded.")
.foregroundStyle(Color.red)
}
if let suggestion = model.quotaUsage.limitIncreaseSuggestion {
Button("Show options") {
suggestion.show()
}
}
}
}
Two things to plan for during the beta. First, the numbers aren’t public and Apple expects to tune them, so test the limit-reached path rather than assuming a budget. Second, a user can reach your app’s daily limit, and once a free allowance runs out, family members draw on one shared pool, so the on-device model is worth keeping as a fallback for tasks it can handle, the same on-device route Google is pushing with offline speech models in AI Edge Gallery. The session describes requests as counted with the user’s iCloud account, while MacRumors describes the free allowance as per device; check Apple’s documentation when the release notes land.
How it compares
Most hosted LLM APIs bill the developer per token and rate-limit the developer’s API key. Apple’s approach puts the cost and the cap on the user instead: the app pays nothing, and when the quota runs out the user waits for the reset or upgrades iCloud+. That makes Apple’s server model cheap to adopt, but it means your feature’s availability depends on how much of that day’s allowance the user has left.