● Covered by 1 source · 2 reportsMedium impact2 neutral
AnyLanguageModel 1.0 Released, Unifying Local and Hosted AI Model Inference on Apple Platforms
Key points
- AnyLanguageModel 1.0 unifies local and hosted AI model inference.
- Supports system models, open-weight models, and major hosted providers.
- New features include tool calling, guided generation, and image support.
- swift-huggingface and swift-xet 1.0 accelerate model downloads.
- AnySpeechRecognizer and AnyDecisionModel are new open-source packages.
- AnySpeechRecognizer extends Apple's speech-to-text API.
- AnySpeechRecognizer adds speaker diarization.
- AnySpeechRecognizer supports NVIDIA's Parakeet and Alibaba's Qwen3-ASR locally on Core ML.
- AnyDecisionModel provides on-device decision-making with typed results.
Unified API for AI Model Inference
AnyLanguageModel 1.0 has been released, offering a shared API for integrating both local and hosted AI models into applications on Apple platforms. This update simplifies the process for developers to select and use different models based on their application's requirements, maintaining a consistent interface.
The core principle of AnyLanguageModel is to use the Foundation Models API with the appropriate model. Developers can switch from `import FoundationModels` to `import AnyLanguageModel` to access this unified functionality.
Expanded Model and Provider Support
The updated framework supports system models, open-weight models through integrations like llama.cpp, MLX, and Core ML, and now includes a wider range of hosted providers. Supported hosted services include OpenAI's Chat Completions and Responses APIs, Anthropic's Messages API, Google's Gemini API, Ollama, and Hugging Face Inference Providers.
AnyLanguageModel also works with any model conforming to Foundation Models' LanguageModel protocol on OS 27, including Core AI models. This expansion allows for greater flexibility in model choice, from on-device execution to cloud-based inference for larger models.
New Features and Capabilities
Several major features have been added over the past year and are included in this release. These include tool calling and guided generation, which are now compatible with every supported provider. Image support allows attaching images to prompts using Foundation Models 27's syntax, including with system models on iOS, macOS, and visionOS 27.
Accompanying AnyLanguageModel, swift-huggingface 1.0 and swift-xet 1.0 have been developed to provide accelerated model downloads from Hugging Face Hub to Swift applications. Hosted providers now also benefit from features like guided generation, tool calling with streaming, and token usage reporting on responses.
Developer Experience Improvements
The update includes improvements for developers, such as API keys now being handled via a closure instead of a direct string. Reasoning blocks are automatically managed when switching between providers, being omitted from requests if the new provider cannot replay them, ensuring compatibility and smooth transitions.
Updates
- AnySpeechRecognizer and AnyDecisionModel are new open-source packages.
- AnySpeechRecognizer extends Apple's speech-to-text API.
- AnySpeechRecognizer adds speaker diarization.
- AnySpeechRecognizer supports NVIDIA's Parakeet and Alibaba's Qwen3-ASR locally on Core ML.
- AnyDecisionModel provides on-device decision-making with typed results.
✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →
The daily brief
One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.
One email a day. Unsubscribe in one click, any time.
Today's brief
Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.
▶ Play today's brief
New every morning, and the back catalogue is archived by date.
Primary sources
GitHub huggingface/blog GitHub huggingface/AnyLanguageModel GitHub huggingface/swift-huggingface GitHub huggingface/swift-xet GitHub apple/coreai-models GitHub mattt/AnySpeechRecognizerHow outlets covered it
New open-source packages, AnySpeechRecognizer and AnyDecisionModel, are now available for Apple platforms. AnySpeechRecognizer extends Apple's speech-to-text API with alternative recognition engines and speaker diarization, while AnyDecisionModel provides on-device decision-making capabilities with typed results. These packages allow developers to use open models from Hugging Face Hub locally on Apple devices.
AnyLanguageModel 1.0 has been released, along with swift-huggingface 1.0 and swift-xet 1.0, providing a unified API for local and hosted AI model inference on Apple platforms. This update allows developers to use a single session code across various models and providers, including system models, open-weight models via llama.cpp, MLX, Core ML, and hosted services like OpenAI, Anthropic, and Google Gemini.