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.

Hello, I have a question regarding hybrid execution for deep learning models on Apple's Neural Engine and CPU. I am aware that setting the precision of some layers to 32-bit allows hybrid execution across both the Neural Engine and the CPU. However, I would like to know if it is possible to achieve the same with 16-bit precision. Is there any specific configuration or workaround to enable hybrid execution in this case? Any guidance or documentation references would be greatly appreciated. Thank you!

I have been working on a small CV program, which uses fine-tuned U2Netp model converted by coremltools 8.3.0 from PyTorch. It works well on my iPhone (with iOS version 18.5) and my Macbook (with MacOS version 15.3.1). But it fails to load after I upgraded Macbook to MacOS version 15.5. I have attached console log when loading this model. Unable to load MPSGraphExecutable from path /Users/yongzhang/Library/Caches/swiftmetal/com.apple.e5rt.e5bundlecache/24F74/E051B28C6957815C140A86134D673B5C015E79A1460E9B54B8764F659FDCE645/16FA8CF2CDE66C0C427F4B51BBA82C38ACC44A514CCA396FD7B281AAC087AB2F.bundle/H14C.bundle/main/main_mps_graph/main_mps_graph.mpsgraphpackage @ GetMPSGraphExecutable E5RT: Unable to load MPSGraphExecutable from path /Users/yongzhang/Library/Caches/swiftmetal/com.apple.e5rt.e5bundlecache/24F74/E051B28C6957815C140A86134D673B5C015E79A1460E9B54B8764F659FDCE645/16FA8CF2CDE66C0C427F4B51BBA82C38ACC44A514CCA396FD7B281AAC087AB2F.bundle/H14C.bundle/main/main_mps_graph/main_mps_graph.mpsgraphpackage (13) Unable to load MPSGraphExecutable from path /Users/yongzhang/Library/Caches/swiftmetal/com.apple.e5rt.e5bundlecache/24F74/E051B28C6957815C140A86134D673B5C015E79A1460E9B54B8764F659FDCE645/16FA8CF2CDE66C0C427F4B51BBA82C38ACC44A514CCA396FD7B281AAC087AB2F.bundle/H14C.bundle/main/main_mps_graph/main_mps_graph.mpsgraphpackage @ GetMPSGraphExecutable E5RT: Unable to load MPSGraphExecutable from path /Users/yongzhang/Library/Caches/swiftmetal/com.apple.e5rt.e5bundlecache/24F74/E051B28C6957815C140A86134D673B5C015E79A1460E9B54B8764F659FDCE645/16FA8CF2CDE66C0C427F4B51BBA82C38ACC44A514CCA396FD7B281AAC087AB2F.bundle/H14C.bundle/main/main_mps_graph/main_mps_graph.mpsgraphpackage (13) Failure translating MIL->EIR network: Espresso exception: "Network translation error": MIL->EIR translation error at /Users/yongzhang/CLionProjects/ImageSimilarity/models/compiled/u2netp.mlmodelc/model.mil:1557:12: Parameter binding for axes does not exist. [Espresso::handle_ex_plan] exception=Espresso exception: "Network translation error": MIL->EIR translation error at /Users/yongzhang/CLionProjects/ImageSimilarity/models/compiled/u2netp.mlmodelc/model.mil:1557:12: Parameter binding for axes does not exist. status=-14 Failed to build the model execution plan using a model architecture file '/Users/yongzhang/CLionProjects/ImageSimilarity/models/compiled/u2netp.mlmodelc/model.mil' with error code: -14.

I am follwing this tutorial: https://apple.github.io/coremltools/docs-guides/source/convert-a-torchvision-model-from-pytorch.html I have obtained simialr result using the python code. However when I view it in Xcode, the preview prediction percentage confidence is way off I suspect it is due the the output of the model, which is in percentage already and in Xcode it multiply 100 again leading to this result. Please give me any feedback to fix this, thank you.

I've built a model using Create ML, but I can't make it, for the love of God, updatable. I can't find any checkbox or anything related. It's an Activity Classifier, if it matters. I want to continue training it on-device using MLUpdateTask, but the model, as exported from Create ML, fails with error: Domain=com.apple.CoreML Code=6 "Failed to unarchive update parameters. Model should be re-compiled." UserInfo={NSLocalizedDescription=Failed to unarchive update parameters. Model should be re-compiled.}

Hey everyone, I want to add an if statement that would do something along the lines of this: if confidence = 100% { } How could I do this? I already have a createML model. Thank you, Oliver

I am using gemini2.5-flash with SwiftUI. How can I receive a response in JSON?

Using highly optimized Metal Shading Language (MSL) code, I pushed the MacBook Air M2 to its performance limits with the deformable_attention_universal kernel. The results demonstrate both the efficiency of the code and the exceptional power of Apple Silicon. The total computational workload exceeded 8.455 quadrillion FLOPs, equivalent to processing 8,455 trillion operations. On average, the code sustained a throughput of 85.37 TFLOPS, showcasing the chip’s remarkable ability to handle massive workloads. Peak instantaneous performance reached approximately 673.73 TFLOPS, reflecting near-optimal utilization of the GPU cores. Despite this intensity, the cumulative GPU runtime remained under 100 seconds, highlighting the code’s efficiency and time optimization. The fastest iteration achieved a record processing time of only 0.051 ms, demonstrating minimal bottlenecks and excellent responsiveness. Memory management was equally impressive: peak GPU memory usage never exceeded 2 MB, reflecting efficient use of the M2’s Unified Memory. This minimizes data transfer overhead and ensures smooth performance across repeated workloads. Overall, these results confirm that a well-optimized Metal implementation can unlock the full potential of Apple Silicon, delivering exceptional computational density, processing speed, and memory efficiency. The MacBook Air M2, often considered an energy-efficient consumer laptop, is capable of handling highly intensive workloads at performance levels typically expected from much larger GPUs. This test validates both the robustness of the Metal code and the extraordinary capabilities of the M2 chip for high-performance computing tasks.

I am running some experiments with WebGPU using the wgpu crate in rust. I have some Buffers already allocated in the GPU. Is it possible to use those already existing buffers directly as inputs to a predict call in CoreML? I want to prevent gpu to cpu download time as much as possible. Or are there any other ways to do something like this. Is this only possible using the latest Tensor object which came out with Metal 4 ?

I watched this year WWDC25 "Read Documents using the Vision framework". At the end of video there is mention of new DetectHandPoseRequest model for hand pose detection in Vision API. I looked Apple documentation and I don't see new revision. Moreover probably typo in video because there is only DetectHumanPoseRequst (swift based) and VNDetectHumanHandPoseRequest (obj-c based) (notice lack of Human prefix in WWDC video) First one have revision only added in iOS 18+: https://developer.apple.com/documentation/vision/detecthumanhandposerequest/revision-swift.enum/revision1 Second one have revision only added in iOS14+: https://developer.apple.com/documentation/vision/vndetecthumanhandposerequestrevision1 I don't see any new revision targeting iOS26+

After a few attempts I end up at the same spot as categorized here by ChatGPT - I get why it feels that way, Rob — honestly, you’ve walked right into the worst possible moment in Apple’s Core ML toolchain evolution. What you’re running into isn’t your fault; Apple really did rip out the ONNX front-end without offering a clean replacement yet. So every tutorial, every script, and even coremltools itself contradicts what’s actually possible right now. Let’s be brutally clear about the state of things (late 2025): Goal Supported by Apple today Practical workaround Convert ONNX → Core ML using coremltools 9 No. The ONNX importer was removed. Use the legacy coremltools 6.x + onnx-coreml stack in a separate Python 3.9/3.10 env. Convert PyTorch → Core ML directly Yes, but only for simple or traced models. T5 still fails because of dynamic control flow. None reliable without rewriting the model. Run ONNX models directly on iOS Partially (via MLC LLM / Metal Performance Shaders). Works, but that’s an entirely different runtime. So: you’re not crazy, and this isn’t wasted effort—you’ve actually proven that your model exports cleanly and that the only missing piece is Apple’s converter. The ONNX file you built is solid. You just need something that can translate it to Core ML. If your goal is simply “get it running on-device”, you still have two viable moves that aren’t rabbit holes: 1. Use MLC LLM / Metal runtime Skip Core ML altogether. Apple’s mlc-llm (and the “MLCChat” pipeline) can execute your ONNX or quantized model directly on the GPU. You keep full fidelity and avoid all these broken converters. 2. Use the legacy Core ML 6.x toolchain in a sandboxed Python 3.10 VM It actually works. You don’t have to downgrade your whole system—just run that specific environment once to emit the .mlpackage, then never touch it again. Nothing else will silently fix itself until Apple re-adds ONNX ingestion in Core ML 10 or a new converter.

Posting a follow up question after the WWDC 2025 Machine Learning AI & Frameworks Group Lab on June 12. In regards to the on-device API of any of the AI frameworks (foundation model, vision framework, ect.), is there a response condition or path where the API outsources it's input to ChatGPT if the user has allowed this like Siri does? Ignore this if it's a no: is this handled behind the scenes or by the developer?

Hi i'm curently crating a model to identify car plates (object detection) i use asitop to monitor my macbook pro and i see that only the cpu is used for the training and i wanted to know why

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'm downloading a fine-tuned model from HuggingFace which is then cached on my Mac when the app first starts. However, I wanted to test adding a progress bar to show the download progress. To test this I need to delete the cached model. From what I've seen online this is cached at /Users/userName/.cache/huggingface/hub However, if I delete the files from here, using Terminal, the app still seems to be able to access the model. Is the model cached somewhere else? On my iPhone it seems deleting the app also deletes the cached model (app data) so that is useful.

Hi, I am modifying the sample camera app that is here: https://developer.apple.com/tutorials/sample-apps/capturingphotos-camerapreview ... In the processPreviewImages, I am using the Vision APIs to generate a segmentation mask for a person/object, then compositing that person onto a different background (with some other filtering). The filtering and compositing is done via CoreImage. At the end, I convert the CIImage to a CGImage then to a SwiftUI Image. When I run it on my iPhone, it works fine, and has not crashed. When I run it on the iPhone with the debugger, it crashes within a few seconds with: EXC_BAD_ACCESS in libRPAC.dylib`std::__1::__hash_table<std::__1::__hash_value_type<long, qos_info_t>, std::__1::__unordered_map_hasher<long, std::__1::__hash_value_type<long, qos_info_t>, std::__1::hash, std::__1::equal_to, true>, std::__1::__unordered_map_equal<long, std::__1::__hash_value_type<long, qos_info_t>, std::__1::equal_to, std::__1::hash, true>, std::__1::allocator<std::__1::__hash_value_type<long, qos_info_t>>>::__emplace_unique_key_args<long, std::__1::piecewise_construct_t const&, std::__1::tuple<long const&>, std::__1::tuple<>>: It had previously been working fine with the debugger, so I'm not sure what has changed. Is there a difference in how the Vision APIs are executed if the debugger is attached vs. not?

I am writing a custom package wrapping Foundation Models which provides a chain-of-thought with intermittent self-evaluation among other things. At first I was designing this package with the command line in mind, but after seeing how well it augments the models and makes them more intelligent I wanted to try and build a SwiftUI wrapper around the package. When I started I was using synchronous generation rather than streaming, but to give the best user experience (as I've seen in the WWDC sessions) it is necessary to provide constant feedback to the user that something is happening. I have created a super simplified example of my setup so it's easier to understand. First, there is the Reasoning conversation item, which can be converted to an XML representation which is then fed back into the model (I've found XML works best for structured input) public typealias ConversationContext = XMLDocument extension ConversationContext { public func toPlainText() -> String { return xmlString(options: [.nodePrettyPrint]) } } /// Represents a reasoning item in a conversation, which includes a title and reasoning content. /// Reasoning items are used to provide detailed explanations or justifications for certain decisions or responses within a conversation. @Generable(description: "A reasoning item in a conversation, containing content and a title.") struct ConversationReasoningItem: ConversationItem { @Guide(description: "The content of the reasoning item, which is your thinking process or explanation") public var reasoningContent: String @Guide(description: "A short summary of the reasoning content, digestible in an interface.") public var title: String @Guide(description: "Indicates whether reasoning is complete") public var done: Bool } extension ConversationReasoningItem: ConversationContextProvider { public func toContext() -> ConversationContext { // <ReasoningItem title="${title}"> // ${reasoningContent} // </ReasoningItem> let root = XMLElement(name: "ReasoningItem") root.addAttribute(XMLNode.attribute(withName: "title", stringValue: title) as! XMLNode) root.stringValue = reasoningContent return ConversationContext(rootElement: root) } } Then there is the generator, which creates a reasoning item from a user query and previously generated items: struct ReasoningItemGenerator { var instructions: String { """ <omitted for brevity> """ } func generate(from input: (String, [ConversationReasoningItem])) async throws -> sending LanguageModelSession.ResponseStream<ConversationReasoningItem> { let session = LanguageModelSession(instructions: instructions) // build the context for the reasoning item out of the user's query and the previous reasoning items let userQuery = "User's query: \(input.0)" let reasoningItemsText = input.1.map { $0.toContext().toPlainText() }.joined(separator: "\n") let context = userQuery + "\n" + reasoningItemsText let reasoningItemResponse = try await session.streamResponse( to: context, generating: ConversationReasoningItem.self) return reasoningItemResponse } } I'm not sure if returning LanguageModelSession.ResponseStream<ConversationReasoningItem> is the right move, I am just trying to imitate what session.streamResponse returns. Then there is the orchestrator, which I can't figure out. It receives the streamed ConversationReasoningItems from the Generator and is responsible for streaming those to SwiftUI later and also for evaluating each reasoning item after it is complete to see if it needs to be regenerated (to keep the model on-track). I want the users of the orchestrator to receive partially generated reasoning items as they are being generated by the generator. Later, when they finish, if the evaluation passes, the item is kept, but if it fails, the reasoning item should be removed from the stream before a new one is generated. So in-flight reasoning items should be outputted aggresively. I really am having trouble figuring this out so if someone with more knowledge about asynchronous stuff in Swift, or- even better- someone who has worked on the Foundation Models framework could point me in the right direction, that would be awesome!

I've built an iOS app with a novel approach to AI safety: a deterministic, pre-inference validation layer called Newton Engine. Instead of relying on the LLM to self-moderate, Newton validates every prompt BEFORE it reaches the model. It uses shape theory and semantic analysis to detect: • Corrosive frames (self-harm language patterns) • Logical contradictions (requests that undermine themselves) • Delegation attempts (asking AI to make human decisions) • Jailbreak patterns (prompt injection, role-play escapes) • Hallucination triggers (requests for fabricated citations) The system achieves a 96% adversarial catch rate across 847 test cases, with zero false positives on benign prompts. Key technical details: • Pure Swift/SwiftUI, no external dependencies • Runs entirely on-device (no server calls for validation) • Deterministic (same input always produces same output) • Auditable (full trace logging for every validation) I'm preparing to submit to the App Store and wanted to ask: Are there specific App Review guidelines I should reference for AI safety claims? Is there interest from Apple in deterministic governance layers for Apple Intelligence integration? Any recommendations for demonstrating safety compliance during review? The app is called Ada, and the engine is open source at: github.com/jaredlewiswechs/ada-newton Happy to share technical documentation or discuss the architecture with anyone interested. See: parcri.net

I’m keep looking around documentation and some sample codes but still haven’t found example of how was used this type of Network Regressor . Does it take some special parameters to perform on ANE , what size,format of DataFrame ?

The developer tutorial for visual intelligence indicates that the method to detect and handle taps on a displayed entity from the Search section is via an "OpenIntent" associated with your entity. However, running this intent executes code from within my app. If I have the perform() method display UI, it always displays UI from within my app. I noticed that the Google app's integration to visual intelligence has a different behavior-- tapping on an entity does not take you to the Google app -- instead, a Webview is presented sheet-style WITHIN the Visual Intelligence environment (see below) How is that accomplished?

In an under-development MacOS & iOS app, I need to identify various measurements from OCR'ed text: length, weight, counts per inch, area, percentage. The unit type (e.g. UnitLength) needs to be identified as well as the measurement's unit (e.g. .inches) in order to convert the measurement to the app's internal standard (e.g. centimetres), the value of which is stored the relevant CoreData entity. The use of NLTagger and NLTokenizer is problematic because of the various representations of the measurements: e.g. "50g.", "50 g", "50 grams", "1 3/4 oz." Currently, I use a bespoke algorithm based on String contains and step-wise evaluation of characters, which is reasonably accurate but requires frequent updating as further representations are detected. I'm aware of the Python SpaCy model being capable of NER Measurement recognition, but am reluctant to incorporate a Python-based solution into a production app. (ref [https://developer.apple.com/forums/thread/30092]) My preference is for an open-source NER Measurement model that can be used as, or converted to, some form of a Swift compatible Machine Learning model. Does anyone know of such a model?