I am trying to build the example on device model example in video WWDC26/241 (What’s new in the Foundation Models framework). I have included the coreai-models package from GitHub but the build still fails with; What’s new in the Foundation Models framework Suggestions appreciated!

I am currently on Apple's Dev Beta V4 for iOS 27. The first version I installed was the Dev Beta V2, I am desperate to try out the new Siri AI Beta, but it's just not installing for me. I have the ability to "turn siri off" then "on again" and find I get the 2024 Apple Intelligence version fine. But if I choose to try out the new AI Beta, I'm left with "Adding support for Siri is in progress. Siri will be unavailable until the update is complete." It's been in that state for over 48 hours in Beta 4 and I'm left with the OLD OLD Siri globe from pre-Apple intelligence. Am I being too keen and just not leaving it long enough? Or is there a genuine issue at Apple's end, in regard to getting the new Siri to actually fully install?

I'm trying to understand the intended relationship between OpenIntent and the new .system.open App Intent schema introduced in iOS 27. From the documentation: OpenIntent (available since iOS 16) is described as an intent that opens an associated item. iOS 27 introduces the .system.open schema, which also appears to represent opening an entity or piece of app content. My questions are: For an app that supports iOS 27+, is .system.open intended to replace OpenIntent, or do the two serve different purposes? For apps that support both iOS 26 and iOS 27+, is the recommended approach to have two structs that implement the same opening logic, one with @AppIntent(schema: .system.open) and the other implementing the OpenIntent protocol? Thanks! References: open protocol OpenIntent

Hello, I'm a EU-based developer. Our app is distributed worldwide. I'd like to clarify the following regarding Siri AI and EU: is Siri AI unavailable to users based in EU, or to apps based in EU. In other words, will my app developed in Europe work with Siri AI for US users? Or the fact that my app is developed in Europe excludes it from compatibility with Siri AI? Kind regards, Bruno

Has anyone else hit this? We have a Mac app that uses FoundationModels with @Generable types for structured output. Starting with macOS 27 beta 5 every guided generation request began logging tokenizer errors and long structured requests slowed from seconds to minutes. Beta 6 and beta 7 both still have it. Filed as FB24310823 on August 11 with a full sysdiagnose and log captures, and we have appended evidence from each beta since. The signature is easy to check. Stream the log while your app generates: log stream --predicate 'subsystem == "com.apple.tokengenerationcore"' --style compact On an affected machine the inference service (TGOnDeviceInferenceProviderService, category guided) prints these two lines in matched pairs, thousands of times: Generated an empty mask at recognizer index N allowedTokenIDs is empty. Something is likely wrong with the tokenizer What we measured on beta 7 today: 9,008 of those pairs in about five and a half minutes of scanning. The errors start about one second into the first request after a fresh app launch, so it needs no warmup. Requests that normally finish in 4 to 12 seconds take 77 to 170 seconds or longer. On beta 5 we measured decode at roughly 0.3 tokens per second on the worst requests. Short requests still finish at normal speed but they emit the same errors while they run, and the quality of the structured content they return is degraded. On betas 5 and 6 we also saw repeated asset release errors for instruct_300m.tokenizer and the instruct_3b tokenizer saying the asset is not marked as in use. For what it is worth, a build that ran clean on beta 4 shows the same behavior on beta 5 and later with no app changes, and the same @Generable schema drives both the fast and the slow requests. But we know that does not rule out something on our side, and we would honestly be happy to learn this is our own bug since that would mean we can fix it. So two questions. Is anyone else seeing this since beta 5? And if you spot something we might be doing wrong on our end, sessions we should be recreating, schema patterns that stress the constrained decoder, anything at all, we would really appreciate the feedback. If it does turn out you are hitting the same thing, a Feedback referencing FB24310823 would help a lot. Thanks!

My intents and entities show up in Shortcuts, and my tests that use App Intents Framework pass. But I can't for the life of me figure out why Siri won't work. I'm trying phrases like "Add to my list in ". All I ever get from Siri is variations of "I can't add items directly to " or "I can't add items to your lists in ". Does anyone see any issues with the following? ( I've left out some of the AppEnum and Entity types for brevity, but these are the main ones) @AppIntent(schema: .reminders.createReminder) struct AddToListIntent { var title: String var list: ListEntity? var note: AttributedString? var isFlagged: Bool? var images: [IntentFile] var tags: Set<String> var urls: [URL] var dueDate: DateComponents? var recurrence: Calendar.RecurrenceRule? var locationTrigger: LocationTriggerEntity? var section: SectionEntity? func perform() async throws -> some ReturnsValue<ReminderEntity> { let newReminder = ReminderEntity(id: "foo", reminder: .init(name: title)) return .result(value: newReminder) } } struct Reminder { var name: String } @AppEntity(schema: .reminders.reminder) struct ReminderEntity { // MARK: Static static let defaultQuery = ReminderEntityQuery() // MARK: Properties let id: String let reminder: Reminder @ComputedProperty(title: "Title") var title: String { reminder.name } var note: AttributedString? { nil } var tags: Set<String> { Set() } var urls: [URL] { [] } var dueDate: DateComponents? { nil } var recurrence: Calendar.RecurrenceRule? { nil } var isCompleted: Bool { false } var isFlagged: Bool? { nil } var creationDate: Date? { nil } var completionDate: Date? { nil } var list: ListEntity var locationTrigger: LocationTriggerEntity? { nil } var displayRepresentation: DisplayRepresentation { .init(title: "\(title)") } // MARK: Query struct ReminderEntityQuery: EntityQuery, EnumerableEntityQuery { func entities(for identifiers: [ReminderEntity.ID]) async throws -> [ReminderEntity] { identifiers.map { .init(id: $0, reminder: .init(name: "Foo")) } } func allEntities() async throws -> [ReminderEntity] { ["foo", "bar", "baz"].map { ReminderEntity(id: $0, reminder: .init(name: $0)) } } } } @AppEntity(schema: .reminders.list) struct ListEntity: AppEntity, IndexedEntity { let id: String let myName: String var name: String { myName } // 3. Define how this entity is displayed to the user in shortcuts/Siri var displayRepresentation: DisplayRepresentation { DisplayRepresentation(title: "\(myName)") } @Property var type: MyListType // 4. Provide a query so the system can resolve specific lists static var defaultQuery = ListEntityQuery() }

Circling back on the LocalLM Lab arc. With v0.7, we've moved from prompt experimentation into real app development on Apple's Foundation Models local AI. The LocalLM Lab SDK lets you build that same on-device model and MCP client this thread has covered directly into your own app, with real tool and data access (Slack, Todoist, GitHub, Notion, Linear, plus Calendar, Reminders, Contacts and Location). And you can ship your app including through the Mac App Store. This is a big improvement over version 0.6, where the localai-cli toolkit needed LocalLM Lab installed and running. On the other hand, the SDK (LocalLMLabSDKCore) doesn't relay through anything; it links FoundationModels and a real MCP client directly into your own binary and is totally self-contained. The example included in the SDK, Plate Today, has actually been built into a sandboxed test app and verified working, with a signed path to a Mac App Store .pkg (Apple Distribution signing + provisioning profile pipeline). That's "verified signable and sandbox-compatible," to be precise. Entitlements (from personal experience: always a complicated topic): com.apple.security.app-sandbox + com.apple.security.network.client for the app itself, plus the standard personal-information entitlements per connector used (com.apple.security.personal-information.calendars, .addressbook, .location) and matching NS*UsageDescription strings in Info.plist. The one worth flagging specifically: the network entitlement is easy to miss and fails silently rather than throwing. Without it, MCP connections and Weather calls just hang with no error surfaced. OAuth handling requires the app delegate callback (application(_:open:)), not SwiftUI's .onOpenURL. Worth knowing before wiring it up if you're SwiftUI-only. Full entitlements list + SDK guide: https://github.com/ancientcomputing/locallm/blob/main/docs/sdk-guide.md Feature page: thisbrain.ai/locallm/sdk.html I hope the availability of the SDK (free, Apache 2.0 license) will give folks further incentive to explore local AI-enabled applications on the Mac. What else would you want to do that the SDK doesn't currently support? File picker? Calendar/Reminders/Contacts edits & writes?

I am developing a third-party macOS application that uses Apple Foundation Models, including Private Cloud Compute (PCC). I would like to confirm the supported distribution requirements for this use case. Specifically: Can a third-party macOS application use Apple Foundation Models / PCC as part of its application functionality? Is PCC usage supported when the macOS application is distributed outside the Mac App Store using Developer ID signing and Apple notarization? Are there any additional entitlements, distribution requirements, or restrictions for PCC when distributing outside the Mac App Store? I intend to use only Apple's documented and supported APIs and will not attempt to bypass PCC availability, quota, entitlement, or other platform restrictions. Thank you.

We found an apparent iOS 27 WorkflowKit bug when implementing: @AppIntent(schema: .photos.editAsset) with an entity conforming to: @AppEntity(schema: .photos.asset) Despite Apple’s general guidance that schema entity types may be renamed, Siri only worked when our entity’s Swift type was named exactly AssetEntity. Controlled on-device results: AssetEntity — works PhotoAssetEntity — fails FooAssetEntity — fails For the failing names, neither the entity query nor perform() was reached. WorkflowKit logged: Failed to retrieve entity metadata Error Domain=WFActionErrorDomain Code=6 Siri responded: Unable to retrieve the data information to process. The generated App Intents metadata was internally consistent, and the issue persisted across clean installs and a device restart. Current workaround: name the .photos.asset entity type exactly AssetEntity. Tested with Xcode 27.0 beta (27A5252f) and iPadOS 27.0 (24A5423a). Filed with Apple as FB24604095 for anyone from Apple investigating this behavior.

My phone no longer shows the waitlist for Siri and has the option to "Try New Siri." I select it, continue, continue and the settings change to "Siri (Beta)" and the waitlist option is no longer there, but when using Siri it's the old pre-Apple Intelligence Siri that activates (little bubble at the bottom) and it does not work. Going to Safari and typing "Siri://" opens the New Siri App, but it says "Siri Update in Progress; Adding support for Siri hasn't completed. Open Settings to check the status." The app does not show up in Spotlight. My phone is done Indexing and all signs point to my phone being enrolled to use the New Siri, but it isn't working at all and still has not shown up. I've tried restarting a few times. Anyone experiencing this too?

Since iOS 18, I have an OpenIntent to open documents. For Siri AI, I understood that I need to annotate the entity with @AppIntent(schema: .system.open) for Siri AI to be able to open documents. This is only supported starting with iOS 27. I tried duplicating the intent (one for iOS 27, one for the other versions), however, Xcode complains and says that only one OpenIntent is possible per target entity. How are we supposed to: support Siri AI "open" functionality preserve functionality for older iOS versions ? Thank you

I’m building Summon (https://github.com/NakliTechie/summon), an open-source native macOS launcher that uses the on-device SystemLanguageModel. Summon creates a fresh LanguageModelSession for each query and attaches only the read-only tools relevant to that query. It currently calls prewarm() after the first keystroke using a temporary session, then creates a different session for generation. The documentation describes prewarm(promptPrefix:) as loading the resources required “for this session.” I would value guidance on four points: Is the prewarming benefit scoped to that exact LanguageModelSession instance? Does a later session using the same SystemLanguageModel receive any benefit? For an ephemeral launcher, is retaining one session preferable to creating a fresh session per query? Which Foundation Models Instrument signal identifies an ineffective prewarm or cache invalidation? Thank You Chirag

I’m trying to understand the intended relationship between IndexedEntity.attributeSet, defaultAttributeSet, and displayRepresentation. For example: struct TrailEntity: IndexedEntity { var displayRepresentation: DisplayRepresentation { DisplayRepresentation( title: "(trail.name)", subtitle: "(trail.location)" ) } var attributeSet: CSSearchableItemAttributeSet { let attributes = CSSearchableItemAttributeSet() attributes.keywords = trail.keywords return attributes } } Should attributeSet instead be initialized with defaultAttributeSet and then have the additional attributes assigned to it? var attributeSet: CSSearchableItemAttributeSet { let attributes = defaultAttributeSet attributes.keywords = trail.keywords return attributes } The documentation says defaultAttributeSet contains values derived from displayRepresentation, but it also describes precedence between displayRepresentation and attributeSet, which suggests Spotlight reads them separately during indexing. So what is the intended pattern? Does overriding attributeSet require including defaultAttributeSet to preserve title/subtitle/image metadata, or is attributeSet only meant for additional Core Spotlight metadata? If the latter, what is the intended use case for overriding or directly using defaultAttributeSet?

I have a photo editing app that owns no photo library. I want a user viewing an image in another app (e.g. Photos) to say "filter this image in MyApp" and have Siri hand that on‑screen image to my intent. Targeting iOS 27. What I've tried, and the result in each case: • App Shortcut + @Parameter var image: IntentFile — Siri resolves my other parameters (a filter AppEnum) by voice, but never binds the image; the run fails. • @AppIntent(schema: .photos.setFilter) with a .photos.asset entity — never routes from Photos. • @AppIntent(schema: .system.open): OpenIntent with a custom AppEntity target — "Open this image in MyApp" just launches the app by name; perform() is never called, and the entity query never runs. My understanding from WWDC26 "Build intelligent Siri experiences with App Schemas" (session 240) and "Discover new capabilities in the App Intents framework" (session 345): • Cross‑app content transfer (Transferable + IntentValueRepresentation) seems limited to system value types (IntentPerson, PlaceDescriptor); IntentFile is not a _SystemIntentValue, so an image can't ride that rail. • Onscreen awareness (NSUserActivity.appEntityIdentifier, View Annotations) appears to expose only the foreground app's own content — which here is Photos, not me. Question: Is there a supported way for a third‑party app to receive another app's on‑screen image (vs. a contact/place) through Siri/App Intents today? If so, which API carries the pixels — an IntentFile parameter, @UnionValue, IntentValueQuery, something else — and what must the source app do to make it available? Or is asking "do X to this image in <third‑party app>" simply not supported yet outside Shortcuts?

皆さん、はじめまして。プログラミングに対する知識がないので、AIを使いながらiPadでTelloのドローンを操縦できるアプリを作っています。このアプリはシミュレーション飛行とドローンの実機飛行ができるアプリにしたいのですが、ドローンの実機飛行を行うためのWi‐Fi接続ができません。（iPadでのシミュレーション飛行が可能です）iPadの基本的な設定はAppleに教えていただき、設定変更を行いましたが、どうしても接続できないので、わかる方法を教えていただけると嬉しいです。 コードを添付いたしますので、ご教示ください。 ドローンアプリコード

I am developing a macOS developer tool that uses Apple Foundation Models, including the Private Cloud Compute (PCC) model. On macOS 27, the Foundation Models CLI provides fm serve, which exposes a local Chat Completions API, including: POST /v1/chat/completions My application communicates with this local API on the user's own Mac to provide agent-style development features. The Foundation Models CLI Legal Notice states: “You are also agreeing to not programmatically access or use Apple models through Apple software or services except as expressly permitted.” I would like to confirm whether using the local API intentionally exposed by fm serve from a third-party macOS application distributed to users is considered an expressly permitted use. The application would: use only the interfaces and endpoints officially exposed by the fm CLI; run fm serve locally on the user's Mac; use the user's own Foundation Models / PCC availability and quota; not bypass quota limits; not use private or undocumented APIs; not reverse engineer Apple services. Is this use of fm serve permitted for a distributed third-party macOS application? If so, are there any additional requirements or restrictions that developers should follow when distributing an application that integrates with fm serve in this way? Thank you.

I would like to understand the foundations of connecting my apps api structure to corporate reporting in regards to textile manufacturing. utilizing the core ML and vision through smartphone to link raw data capture and strategic execution.

I'm looking for a supported way to read intermediate tensors from a model executing on the Apple Neural Engine - specifically the output of an individual node in the compiled graph, rather than only the final output. What I'm trying to do: validate a from-weights reimplementation of a model against the real thing, layer by layer. Comparing only the final output tells me the reimplementation is wrong but not where; a per-layer comparison would localise it immediately. What I've established so far: A compiled ANE program can be executed unprivileged through the public graph API, and the final readout matches, so the execution path itself is reachable. Intermediate activations don't appear in host memory during normal operation, which is expected since the scheduler keeps them in accelerator-local storage. Requesting a per-node output appears to hit a kernel-side check that an ordinary process doesn't satisfy. Questions: Is there a supported API for retrieving per-node outputs from an ANE-scheduled graph - a debug or instrumentation mode, an Instruments template, or a Core ML compute-plan facility that surfaces them? Failing that, is there a supported way to make a specific node materialise its output to a host-visible buffer - for example by splitting the graph, marking an intermediate tensor as a model output, or compiling with that node as a terminal operation? I'm aware this may change scheduling and defeat the purpose, but I'd like to know whether it's the intended approach. If neither exists, is that a deliberate design boundary rather than a gap? A clear "no" is a useful answer and I'll stop looking. I'm not asking about any particular shipped model, and this isn't a request to bypass anything - the question is whether the platform exposes per-node observability for ANE execution at all, and if so what the supported entry point is. Thanks.

I’m evaluating a third-party Swift-based “orchestration layer” for enterprise AI workloads on Apple Silicon — it claims to handle job queuing, scheduling, batching, memory management, monitoring, auditing, and failover on top of on-device inference. Given the Core AI framework’s device-specialization step and InferenceFunction pipeline (and MLX’s unified-memory model), how much of this kind of orchestration is already handled natively versus something a developer would still need to build themselves? Specifically: 1. Does Core AI’s inference pipeline provide any built-in job queuing/batching across multiple concurrent requests, or is that entirely app-side? 2. Is there native failover/monitoring tooling for on-device inference, or would a developer need to build that themselves (e.g., via os_log, MetricKit, custom retry logic)? 3. For memory management across CPU/GPU/ANE, does unified memory in MLX/Core AI eliminate most of the manual management a custom orchestration layer would otherwise need to solve? Trying to understand what’s genuinely differentiated in a third-party layer versus what Apple’s stack already provides out of the box. Appreciate any insight from folks who’ve built with Core AI/MLX in production.

import Playgrounds import FoundationModels #Playground { do { let session = LanguageModelSession() let response = try await session.respond( to: "Explain SwiftUI in one sentence." ) print(response.content) } catch { print("Error: \(error)") } }``` I tested Foundation Models with this simple code, and it generated this error: "Error Domain=ModelManagerServices.ModelManagerError Code=1026 \"(null)\" UserInfo={NSMultipleUnderlyingErrorsKey=(\n)}" I tried restarting my Mac and Apple Intelligence, but that didn't work. What did work was updating Xcode and the simulators to the latest possible version.