At WWDC25 we launched a new type of Lab event for the developer community - Group Labs. A Group Lab is a panel Q&A designed for a large audience of developers. Group Labs are a unique opportunity for the community to submit questions directly to a panel of Apple engineers and designers. Here are the highlights from the WWDC25 Group Lab for Machine Learning and AI Frameworks. What are you most excited about in the Foundation Models framework? The Foundation Models framework provides access to an on-device Large Language Model (LLM), enabling entirely on-device processing for intelligent features. This allows you to build features such as personalized search suggestions and dynamic NPC generation in games. The combination of guided generation and streaming capabilities is particularly exciting for creating delightful animations and features with reliable output. The seamless integration with SwiftUI and the new design material Liquid Glass is also a major advantage. When should I still bring my own LLM via CoreML? It's generally recommended to first explore Apple's built-in system models and APIs, including the Foundation Models framework, as they are highly optimized for Apple devices and cover a wide range of use cases. However, Core ML is still valuable if you need more control or choice over the specific model being deployed, such as customizing existing system models or augmenting prompts. Core ML provides the tools to get these models on-device, but you are responsible for model distribution and updates. Should I migrate PyTorch code to MLX? MLX is an open-source, general-purpose machine learning framework designed for Apple Silicon from the ground up. It offers a familiar API, similar to PyTorch, and supports C, C++, Python, and Swift. MLX emphasizes unified memory, a key feature of Apple Silicon hardware, which can improve performance. It's recommended to try MLX and see if its programming model and features better suit your application's needs. MLX shines when working with state-of-the-art, larger models. Can I test Foundation Models in Xcode simulator or device? Yes, you can use the Xcode simulator to test Foundation Models use cases. However, your Mac must be running macOS Tahoe. You can test on a physical iPhone running iOS 18 by connecting it to your Mac and running Playgrounds or live previews directly on the device. Which on-device models will be supported? any open source models? The Foundation Models framework currently supports Apple's first-party models only. This allows for platform-wide optimizations, improving battery life and reducing latency. While Core ML can be used to integrate open-source models, it's generally recommended to first explore the built-in system models and APIs provided by Apple, including those in the Vision, Natural Language, and Speech frameworks, as they are highly optimized for Apple devices. For frontier models, MLX can run very large models. How often will the Foundational Model be updated? How do we test for stability when the model is updated? The Foundation Model will be updated in sync with operating system updates. You can test your app against new model versions during the beta period by downloading the beta OS and running your app. It is highly recommended to create an "eval set" of golden prompts and responses to evaluate the performance of your features as the model changes or as you tweak your prompts. Report any unsatisfactory or satisfactory cases using Feedback Assistant. Which on-device model/API can I use to extract text data from images such as: nutrition labels, ingredient lists, cashier receipts, etc? Thank you. The Vision framework offers the RecognizeDocumentRequest which is specifically designed for these use cases. It not only recognizes text in images but also provides the structure of the document, such as rows in a receipt or the layout of a nutrition label. It can also identify data like phone numbers, addresses, and prices. What is the context window for the model? What are max tokens in and max tokens out? The context window for the Foundation Model is 4,096 tokens. The split between input and output tokens is flexible. For example, if you input 4,000 tokens, you'll have 96 tokens remaining for the output. The API takes in text, converting it to tokens under the hood. When estimating token count, a good rule of thumb is 3-4 characters per token for languages like English, and 1 character per token for languages like Japanese or Chinese. Handle potential errors gracefully by asking for shorter prompts or starting a new session if the token limit is exceeded. Is there a rate limit for Foundation Models API that is limited by power or temperature condition on the iPhone? Yes, there are rate limits, particularly when your app is in the background. A budget is allocated for background app usage, but exceeding it will result in rate-limiting errors. In the foreground, there is no rate limit unless the device is under heavy load (e.g., camera open, game mode). The system dynamically balances performance, battery life, and thermal conditions, which can affect the token throughput. Use appropriate quality of service settings for your tasks (e.g., background priority for background work) to help the system manage resources effectively. Do the foundation models support languages other than English? Yes, the on-device Foundation Model is multilingual and supports all languages supported by Apple Intelligence. To get the model to output in a specific language, prompt it with instructions indicating the user's preferred language using the locale API (e.g., "The user's preferred language is en-US"). Putting the instructions in English, but then putting the user prompt in the desired output language is a recommended practice. Are larger server-based models available through Foundation Models? No, the Foundation Models API currently only provides access to the on-device Large Language Model at the core of Apple Intelligence. It does not support server-side models. On-device models are preferred for privacy and for performance reasons. Is it possible to run Retrieval-Augmented Generation (RAG) using the Foundation Models framework? Yes, it is possible to run RAG on-device, but the Foundation Models framework does not include a built-in embedding model. You'll need to use a separate database to store vectors and implement nearest neighbor or cosine distance searches. The Natural Language framework offers simple word and sentence embeddings that can be used. Consider using a combination of Foundation Models and Core ML, using Core ML for your embedding model.

Hi there, I have a custom keypoint detection model and want to use it via vision's CoremlRequest API. Here's some complication for input and output: For input My model expect 512x512 a image. Which would be resized and padded from a 1920x1080 frame. I use the .scaleToFit option, but can I also specify the color used for padding? For output: My model output a CoreMLFeatureValueObservation, can I have it output in a format vision recognizes? such as joints/keypoints If my model is able to output in a format vision recognizes, would it take care to restoring the coordinates back to the original frame? (undo the padding) If not, how do I restore it from .scaletofit option? Best,

I'm adding Visual Intelligence support to my app, and now want to add a Tip using TipKit to guide users to this feature from within my app. I want to add a Rule to my Tip which will only show this Tip on devices where Visual Intelligence is supported (ex. not iPhone 14 Pro Max). What is the best way for me to determine availability to set this TipKit rule? Here's the documentation I'm following for Visual Intelligence: https://developer.apple.com/documentation/visualintelligence/integrating-your-app-with-visual-intelligence

Are there any guidelines for using Foundation Models To generate text for users in response to some canned queries? Should we use a special icon or text to let the user know that Apple Intelligence is generating the text? Should there be a disclaimer like, Apple Intelligence can make mistakes, please check for accuracy, etc?

I've created a "Transfer Learning BERT Embeddings" model with the default "Latin" language family and "Automatic" Language setting. This model performs exceptionally well against the test data set and functions as expected when I preview it in Create ML. However, when I add it to the Xcode project of the application to which I am deploying it, I am getting runtime errors that suggest it can't find the embedding resources: Failed to locate assets for 'mul_Latn' - '5C45D94E-BAB4-4927-94B6-8B5745C46289' embedding model Note, I am adding the model to the app project the same way that I added an earlier "Maximum Entropy" model. That model had no runtime issues. So it seems there is an issue getting hold of the embeddings at runtime. For now, "runtime" means in the Simulator. I intend to deploy my application to iOS devices once GM 26 is released (the app also uses AFM). I'm developing on Tahoe 26 beta, running on iOS 26 beta, using Xcode 26 beta. Is this a known/expected issue? Are the embeddings expected to be a resource in the model? Is there a workaround? I did try opening the model in Xcode and saving it as an mlpackage, then adding that to my app project, but that also didn't resolve the issue.

Hello everyone, I have a visual convolutional model and a video that has been decoded into many frames. When I perform inference on each frame in a loop, the speed is a bit slow. So, I started 4 threads, each running inference simultaneously, but I found that the speed is the same as serial inference, every single forward inference is slower. I used the mactop tool to check the GPU utilization, and it was only around 20%. Is this normal? How can I accelerate it?

When the @Generable is applied toward a Swift struct declared within another struct, and when said nested struct is defined as the type of one of the properties of another @Generable type, which is in turn defined as the output format of Foundation Model session, Foundation Model can stuck in a loop trying to create a infinitely nested response, until the context window limit exceeded error is triggered. I have filed feedback FB19987191 with a demo project. Is this expected behavior?

Lately I am getting this error. GenerativeModelsAvailability.Parameters: Initialized with invalid language code: en-GB. Expected to receive two-letter ISO 639 code. e.g. 'zh' or 'en'. Falling back to: en Does anyone know what this is and how it can be resolved. The error does not crash the app

Hi, Are there rules around using Foundation Models: In a background task/session? Concurrently, i.e. a bunch simultaneously using Swift Concurrency? I couldn't find this in the docs (sorry if I missed it) so wondering what's supported and what the best practice is here. In case it matters, my primary platform is Vision Pro (so, M2).

I have integrated Apple’s Foundation Model into my iOS application. As known, Foundation Model is only supported starting from iOS 26 on compatible devices. To maintain compatibility with older iOS versions, I wrapped the API calls with the condition if #available(iOS 26, *). The application works normally on an iPad running iOS 18 and on a Mac running macOS 26. However, when running the same build on a MacBook Air M1 (macOS 15) through iPad app compatibility, the app crashes immediately upon launch. The main issue is that I cannot debug directly on macOS 15, since the app can only be built on macOS 26 with Xcode beta. I then have to distribute it via TestFlight and download it on the MacBook Air M1 for testing. This makes identifying the detailed cause of the crash very difficult and time-consuming. Nevertheless, I have confirmed that the crash is caused by the Foundation Model APIs.

Foundation Models are driving me up the wall. My use case: A news app - I want to summarize news articles. Sounds like a perfect use for the added-in-beta-5 "no guardrails" mode for text-to-text transformations... ... and it's true, I don't get guardrails exceptions anymore but now, the model itself frequently refuses to summarize stuff which in a way is even worse as I have to parse the output text to figure out if it failed instead of getting an exception. I mostly worked that out with my system instructions but still, the refusing to summarize makes it really tough to use. I instructed the model to tell me why it failed if that happens. Examples of various refusals for news articles from major sources: "The article mentions "Visual Lookup" but does not provide details about how it integrates with iOS 26." "The article includes unsafe content regarding a political figure's potential influence over the Federal Reserve board, which is against my guidelines." "the article contains unsafe content." "The article is biased and opinionated and focuses on the author's opinion." (this is despite the instructions specifically asking for a neutral summary - I am asking it to not use bias in the output but it still refuses) I have tons of these. Note that if I don't use the "no guardrails" mode and use a Generable instead, some of these work fine so right now I have to do two passes on much of the content since I never know which one will work. Having a "summary mode" that often refuses to summarize current news articles (the world is not a great place, some of these stories are a bummer) is near worthless.

JAX Metal shows 55x slower random number generation compared to NVIDIA CUDA on equivalent workloads. This makes Monte Carlo simulations and scientific computing impractical on Apple Silicon. Performance Comparison NVIDIA GPU: 0.475s for 12.6M random elements M1 Max Metal: 26.3s for same workload Performance gap: 55x slower Environment Apple M1 Max, 64GB RAM, macOS Sequoia Version 15.6.1 JAX 0.4.34, jax-metal latest Backend: Metal Reproduction Code import time import jax import jax.numpy as jnp from jax import random key = random.PRNGKey(42) start_time = time.time() random_array = random.normal(key, (50000, 252)) duration = time.time() - start_time print(f"Duration: {duration:.3f}s")

I've tried creating a Lora adapter using the example dataset, scripts as part of the adapter_training_toolkit_v26_0_0 (last available) on MacOs 26 Beta 6. import SwiftUI import FoundationModels import Playgrounds #Playground { // The absolute path to your adapter. let localURL = URL(filePath: "/Users/syl/Downloads/adapter_training_toolkit_v26_0_0/train/test-lora.fmadapter") // Initialize the adapter by using the local URL. let adapter = try SystemLanguageModel.Adapter(fileURL: localURL) // An instance of the the system language model using your adapter. let customAdapterModel = SystemLanguageModel(adapter: adapter) // Create a session and prompt the model. let session = LanguageModelSession(model: customAdapterModel) let response = try await session.respond(to: "hello") } I get Adapter assets are invalid error. I've added the entitlements Is adapter_training_toolkit_v26_0_0 up to date?

Was just wondering why the foundation model documentation is no longer available, thanks! https://developer.apple.com/documentation/FoundationModels

Due to our min iOS version, this is my first time using .xcstrings instead of .strings for AppShortcuts. When using the migrate .strings to .xcstrings Xcode context menu option, an .xcstrings catalog is produced that, as expected, has each invocation phrase as a separate string key. However, after compilation, the catalog changes to group all invocation phrases under the first phrase listed for each intent (see attached screenshot). It is possible to hover in blank space on the right and add more translations, but there is no 1:1 key matching requirement to the phrases on the left nor a requirement that there are the same number of keys in one language vs. another. (The lines just happen to align due to my window size.) What does that mean, practically? Do all sub-phrases in each language in AppShortcuts.xcstrings get processed during compilation, even if there isn't an equivalent phrase key declared in the AppShortcut (e.g., the ja translation has more phrases than the English)? (That makes some logical sense, as these phrases need not be 1:1 across languages.) In the AppShortcut declaration, if I delete all but the top invocation phrase, does nothing change with Siri? Is there something I'm doing incorrectly? struct WatchShortcuts: AppShortcutsProvider { static var appShortcuts: [AppShortcut] { AppShortcut( intent: QuickAddWaterIntent(), phrases: [ "\(.applicationName) log water", "\(.applicationName) log my water", "Log water in \(.applicationName)", "Log my water in \(.applicationName)", "Log a bottle of water in \(.applicationName)", ], shortTitle: "Log Water", systemImageName: "drop.fill" ) } }

Hi, I am a new IOS developer, trying to learn to integrate the Apple Foundation Model. my set up is: Mac M1 Pro MacOS 26 Beta Version 26.0 beta 3 Apple Intelligence & Siri --> On here is the code, func generate() { Task { isGenerating = true output = "⏳ Thinking..." do { let session = LanguageModelSession( instructions: """ Extract time from a message. Example Q: Golfing at 6PM A: 6PM """) let response = try await session.respond(to: "Go to gym at 7PM") output = response.content } catch { output = "❌ Error:, \(error)" print(output) } isGenerating = false } and I get these errors guardrailViolation(FoundationModels.LanguageModelSession.GenerationError.Context(debugDescription: "Prompt may contain sensitive or unsafe content", underlyingErrors: [Asset com.apple.gm.safety_embedding_deny.all not found in Model Catalog])) Can you help me get through this?

I'm a bit new to the LLM stuff and with Foundation Models. My understanding is that there is a token limit of around 4K. I want to process the contents of files which may be quite large. I first tried going the Tool route but that didn't work out so I then tried manually chunking the text to keep things under the limit. It mostly works except that every now and then it'll exceed the limit. This happens even when the chunks are less than 100 characters. Instructions themselves are about 500 characters but still overall, well below 1000 characters per prompt, all told, which, in my limited understanding, should not result in 4K tokens being parsed. Any ideas on what is going on here?

I keep getting the error “An unsupported language or locale was used.” Is there any documentation that specifies the accepted languages or locales in Foundation model?

Hello, I am testing the sample project provided here: Bringing advanced speech-to-text capabilities to your app. On both macOS 26.0 beta and iOS 26.0 beta, the app crashes immediately on launch with a dyld "Symbol not found" error related to FoundationModels.framework. It feels like this may be related to testing primarily on newer Apple Silicon devices, as I am seeing consistent crashes on an Intel MacBook and on an older iPhone device. I would appreciate any insight, confirmation, or guidance on whether this is a known limitation or if there is a workaround. Is it planned to be resolved soon? Environment macOS: Device: MacBook Pro (Intel) Processor: 2 GHz Quad-Core Intel Core i5 Graphics: Intel Iris Plus Graphics 1536 MB Memory: 16 GB 3733 MHz LPDDR4X OS: macOS Tahoe Version 26.0 Beta (25A5338b) iOS: Device: iPhone 11 Model Number: MHDD3HN/A OS: iOS 26.0 Xcode: Version: 26.0 beta 3 (17A5276g) Crash (macOS) Abort signal received. Excerpt from crash dump: dyld`__abort_with_payload: 0x7ff80e3ad4a0 <+0>: movl $0x2000209, %eax 0x7ff80e3ad4a5 <+5>: movq %rcx, %r10 0x7ff80e3ad4a8 <+8>: syscall -> 0x7ff80e3ad4aa <+10>: jae 0x7ff80e3ad4b4 Console: dyld[9819]: Symbol not found: _$s16FoundationModels20LanguageModelSessionC5model10guardrails5tools12instructionsAcA06SystemcD0C_AC10GuardrailsVSayAA4Tool_pGAA12InstructionsVSgtcfC Referenced from: /Users/userx/Library/Developer/Xcode/DerivedData/SwiftTranscriptionSampleApp-*/Build/Products/Debug/SwiftTranscriptionSampleApp.app/Contents/MacOS/SwiftTranscriptionSampleApp.debug.dylib Expected in: /System/Library/Frameworks/FoundationModels.framework/Versions/A/FoundationModels Crash (iOS) Abort signal received. Excerpt from crash dump: dyld`__abort_with_payload: 0x18f22b4b0 <+0>: mov x16, #0x209 0x18f22b4b4 <+4>: svc #0x80 -> 0x18f22b4b8 <+8>: b.lo 0x18f22b4d8 Console dyld[2080]: Symbol not found: _$s16FoundationModels20LanguageModelSessionC5model10guardrails5tools12instructionsAcA06SystemcD0C_AC10GuardrailsVSayAA4Tool_pGAA12InstructionsVSgtcfC Referenced from: /private/var/containers/Bundle/Application/.../SwiftTranscriptionSampleApp.app/SwiftTranscriptionSampleApp.debug.dylib Expected in: /System/Library/Frameworks/FoundationModels.framework/FoundationModels Question Is this crash expected on Intel Macs and older iPhone models with the beta SDKs? Is there an official statement on whether macOS 26.x releases support Intel, or it exists only until macOS 26.1? Any suggested workarounds for testing this sample project on current hardware? Is this a known limitation for the 26.0 beta, and if so, should we expect a fix in 26.0 or only in subsequent releases? Attaching screenshots for reference. Thank you in advance.

Dear Apple Foundation Models Development Team, I am a developer integrating Apple Foundation Models (AFM) into my app and encountered the exceededContextWindowSize error when exceeding the 4096-token limit. Proposal: I suggest Apple develop a tool to estimate the token count of a prompt before sending it to the model. This tool could be integrated into FoundationModels Framework for ease of use. Benefits: A token estimation tool would help developers manage the context window limit and optimize performance. I hope Apple considers this proposal soon. Thank you!

Hello, My app fully relies on the new Foundation Models. Since Foundation Models require Apple Intelligence, I want to ensure that only devices capable of running Apple Intelligence can install my app. When checking the UIRequiredDeviceCapabilities property for a suitable value, I found that iphone-performance-gaming-tier seems the closest match. Based on my research: On iPhone, this effectively limits installation to iPhone 15 Pro or later. On iPad, it ensures M1 or newer devices. This exactly matches the hardware requirements for Apple Intelligence. However, after setting iphone-performance-gaming-tier, I noticed that on iPad, Game Mode (Game Overlay) is automatically activated, and my app is treated as a game. My questions are: Is there a more appropriate UIRequiredDeviceCapabilities value that would enforce the same Apple Intelligence hardware requirements without triggering Game Mode? If not, is there another way to restrict installation to devices meeting Apple Intelligence requirements? Is there a way to prevent Game Mode from appearing for my app while still using this capability restriction? Thanks in advance for your help.