There seems to be a lack of clarity about how this Siri AI rollout is working. Does the indexing have anything to do with getting taken off the waitlist, or are they completely separate? Also, what is the expected wait time and has anyone gotten the new Siri after the initial wave of approvals (after 4 hours after the keynote)? The waitlist is extremely slow compared to the initial Apple Intelligence waitlist from a couple of years ago.

We're doing some local LLM inference experiments on Apple Silicon and have observed something that seems counterintuitive. In several cases, MoE models significantly outperform dense models despite having similar total parameter counts. Examples (simplified): Dense model: ~30B parameters MoE model: ~30B total parameters, ~3B active parameters On Apple Silicon, the MoE model consistently achieves higher decode throughput even after accounting for routing overhead. A few hypotheses we're considering: Active parameter count appears to matter more than total parameter count for decode throughput. Memory traffic may dominate M=1 autoregressive decode, making sparse activation more important than expected. Expert matrix geometry might matter as much as parameter count. Some MoE designs appear to produce GPU-friendly GEMV shapes while others do not. Quantization layout and memory alignment seem to have surprisingly large effects on practical throughput. What I'm curious about is whether others have observed similar behavior on Apple Silicon specifically. Has anyone profiled decode throughput across: dense models large-expert MoE many-small-expert MoE and identified which hardware characteristics are actually driving the difference? I'm particularly interested in observations from Metal profiling rather than benchmark leaderboards.

We're doing some research on Apple Silicon inference runtimes and trying to understand the practical synchronization boundary of Apple GPUs. We are not asking about threadgroup barriers (those are documented), but about device-scope synchronization patterns built from atomics. What we've observed: Device-scope atomics are available. It is possible to build global counters and persistent-thread style coordination structures. However, we cannot find any documented guarantee regarding: threadgroup co-residency, global forward progress, occupancy-bounded synchronization safety. In our experiments, synchronization schemes that rely on all threadgroups making progress eventually can become unreliable, while strictly local producer/consumer handoff patterns appear much more robust. Questions: Does Metal provide any documented forward-progress guarantees across threadgroups beyond what is explicitly stated in the Metal specification? Is there any recommended pattern for implementing long-lived producer/consumer GPU pipelines without relying on global synchronization assumptions? For Apple GPUs specifically, should developers assume that occupancy-bounded global synchronization is unsupported unless explicitly provided by the API? We are not looking for undocumented implementation details, only for guidance on what assumptions are safe for production systems. Thanks.

Subject: Inquiry Regarding Architectural Overhead and Buffer Access in the Push to Talk Framework for Real-Time Core ML Blind Source Separation Dear Apple Engineering Team, We are currently developing an Apple-native communication platform that utilizes the Push to Talk framework alongside Core ML to handle real-time, on-device audio processing. We are working to resolve the issue of single-channel, co-channel interference (overlapping voice streams) directly on the edge. Our current challenge lies in the pipeline latency and background lifecycle constraints when intercepting incoming audio buffers. To cleanly separate overlapping voices before they hit the audio output mixer, we need to process the raw PCM data immediately upon arrival. Could you please provide guidance on the following architectural questions: Low-Latency Buffer Interception: What is the recommended design pattern within the PTChannelManagerDelegate flow to pass raw incoming audio buffers directly to a Core ML model running on the Apple Neural Engine (ANE) before the system routes them to AVAudioEngine for playback? Background Thread Management: Given the strict background execution boundaries enforced by the Push to Talk framework, how can we best optimize thread scheduling to ensure our speech separation model completes its execution without triggering an OS background processing timeout or process termination? Dynamic UI Manifestation: Once a combined audio stream is separated into two clean, distinct voice vectors on-device, what is the best approach for registering multiple PTParticipant states simultaneously so that the native system UI (like the Dynamic Island) accurately reflects both speakers? Thank you for your time, insights, and continued support of developer innovation within the iOS and iPadOS ecosystems. Best regards, Ken Zakreski Founder, Marine Link Pro

I have been waiting for both of these going on 24 hours with nothing happening

Since i got the iPadOS27 update signed in the waitlist and haven’t got any new yet

I just watched the "Build Intelligent Siri experiences with App Schemas" and I'm confused about how to integrate my app with the new Apple Intelligence + Siri in iOS27. I think it mentions specifically that Siri needs to adopt App Schemas, and that just adopting App Intents in my app isn't enough for it to integrate with the new Siri. Is that correct? The 'schemas' seem to be a narrow set of specific activities. What if my app's actions (or intents) don't match closely with it? For example, in my app, I have entities like Tags and Contacts. I can 'create tag' as well as 'add tags to a contact' as 2 different App intents. If I'm using just App Intents on their own, would these not map to the new Siri? I can also add a 'task' to a 'contact'. Would that possibly work with Siri? The videos just don't seem to make an effort to explain what is and what isn't possible.

Is the foundation model planned to be a part of the OS, or will we need to keep checking for its availability in our iOS apps? It would be cool to develop an app that heavily uses it as its core, especially given its privacy and offline capabilities.

Hello, I have an iPhone 15 Pro running iOS 27 Developer Beta. The “Apple Intelligence & Siri” menu does not appear in Settings. I only have the regular Siri menu, and Apple Intelligence features such as Writing Tools are unavailable. I have already: Set my iPhone language to English (US) Set my region to United States Updated to iOS 27 Developer Beta Restarted the device What confuses me is that this same iPhone previously belonged to my father, and I clearly remember Apple Intelligence being available on this device before I signed in with my own Apple Account. After switching to my account, Apple Intelligence disappeared completely. My model number is MV923CH/A. Has anyone experienced this or knows what might be causing it? Thank you.

Hi! From my past experience, I know that foundation models from last year have really small context window and this year with the new foundation models. I suspect the context window gets much larger, but if the new Siri AI is coming only two iPhone 17. Does that mean my iPhone 16 will not get that new foundation model models, bigger context.

If some Siri AI features work on cloud why not for older device? Apple Devs mentioned that the Siri AI is available for iPhone 15 and later devices and we can use iCloud+ Subscription to use it more on Apple’s Private Cloud Compute (PCC). So if it can run some functions on Private Cloud why not for people with older device? Or will Apple make it compatible with older devices on upcoming beta updates? Also, I noticed that other Phone companies have these features in phones that are way less powerful than iPhone 11 (Which is a oldest device support iOS 27) I tested some ai models from 3rd party apps that are less than 300MB and it worked very well than the old siri so can We add some Siri features to older iPhone? So we can avoid people from switching to other phones. Because I don’t want my friends leaving iPhone.

It has been over 3 years since you chose, to save Apple money, you changes voice to text, and it is worthless. Do you plan on fixing this, or going back to paying for cloud service, instead of being cheap, sacrificing our experience so you can save money your trillion dollar company does not need.

: Hello, I’m seeking clarification on whether Apple provides any framework or API that enables deep integration between Siri and advanced AI assistants (such as ChatGPT), including system-level functions like voice interaction, navigation, cross-platform syncing, and operational access similar to Siri’s own capabilities. If no such option exists today, I would appreciate guidance on the recommended path or approved third-party solutions for building a unified, voice-first experience across Apple’s ecosystem. Thank you for your time and insight.

Hi everyone, I can’t find a way to request the Private Cloud Compute entitlement in the Capability Requests. However, in the WWDC26 video “Build with the new Apple Foundation Model on Private Cloud Compute” it says you can request it now. I’m hoping I’m missing something. Any help would be great!

Sorry I thought it would tell you in the description or in the forum but when I went to try to activate new Siri it says that I have to be on a waitlist and from all that I've checked I'm an Apple developer so I'm wondering if anybody else is waiting or if the system is so busy today. If anybody else is having this issue please let me know.

My app uses ImageCreator to allow customers to generate images on-device, and provides a custom experience for doing so. I can see that ImageCreator is now being deprecated without a replacement for programatic image generation. Why has the ability to generate images programatically been removed?

Hello. I have a problem with the FoundationModels adapter and the Apple-hosted managed asset pack via TestFlight. I have created an adapter that works fine locally by creating a model via (fileURL: URL) on a real device, but I cannot create a model using background assets by downloading the adapter via TestFlight. Every time I try to get an adapter, the creation of the adapter is interrupted by the compatibleAdapterNotFound error. The aar. archive i created using a special command - xcrun ba-package foundation-models package --adapter-path aurelius1.fmadapter --asset-pack-id fmadapter-aurelius1-9799725 --output-path ./aurelius1.aar --platforms iOS --on-demand\ after that, I replaced "OnDemand": null with "OnDemand": {} in the manifest so that the Transporter could send my archive to the App Store Connect. I followed all the recommendations in this topic - https://origin-devforums.apple.com/forums/thread/823148 ...but unfortunately unsuccessfully I would appreciate any help in solving this problem. here is the code that I use in my app -

Can't able to run the Create ML for training and I upgraded to MacOS 26.3 beta and I have tried older and newer

Hi! I noticed that on my father's M1 Max MacBook Pro (64gb ram) there's an option for style transfer which I don't see on my M1 MacBook Air (16gb ram). I am running macOS Tahoe and he is running macOS Sequoia.

I see a linear slow down as more trackers are added in the loop below. According to the WWDC video I was hoping that performAll run all these request in parallel, but apparently not for TrackObjectRequest.... #if DEBUG print("Frame processor: \(requests.count) tracking requests") var observationCount: Int = 0 let trackingStart = Date() #endif for await observation in handler.performAll(requests) { if case .trackObject(let request, let trackedBlock) = observation { #if DEBUG observationCount += 1 #endif guard let trackedBlock = trackedBlock, trackedBlock.confidence <= FrameProcessingThresholds.blockTrackedConfidenceThreshold else { // lost track of the block, remove the tracker blockTrackers.removeValue(forKey: request) continue } trackedBlocks[blockTrackers[request]!] = trackedBlock.boundingBox } } #if DEBUG print("Frame processor: \(observationCount) observations") print("Frame processor: tracker took \(Date().timeIntervalSince(trackingStart)) seconds") #endif