AnyLanguageModel 1.0 unifies local and cloud models behind one API on Apple platforms
Hugging Face’s Swift package provides a shared session API compatible with Apple Foundation Models, allowing apps to switch among the system model, local open-weight models, and hosted providers.
AT A GLANCE
Key points
- Drop-in compatible API for Apple Foundation Models.
- Supports local models through llama.cpp, MLX, and Core ML.
- Works with major hosted APIs and Hugging Face Inference Providers.
- Tool calling and guided generation are available across providers.
What is changing?
Hugging Face released AnyLanguageModel 1.0 on October 20. The open-source Swift package provides an API compatible with Apple’s Foundation Models framework. By changing only the import statement, developers can use the system model, local models through llama.cpp, MLX, and Core ML, or hosted models with the same session code.
In version 1.0, tool calling and guided generation work across all providers. Image prompts extend to the system model on iOS, macOS, and visionOS 27. Hosted options include OpenAI, Anthropic, Gemini, Ollama, Private Cloud Compute, and Hugging Face Inference Providers, and the package also runs on Linux.
This approach can make it easier for Apple apps to choose between on-device privacy and offline operation or the capacity of cloud models on a task-by-task basis. Provider differences, key management, device memory, and model download sizes still need production testing.
Explore the original sources
Source date: 20 October 2026.
Published by AI Radar: 21 Oct 2026, 00:01 (Istanbul, UTC+3).
FOLLOW THE TOPIC
Topics and tags
Share your discovery.
Send this story to a friend.
Reader comments
What do you think about this development? Share your experience or question. Comments are published after editorial approval.
Your name/nickname and comment become public if approved. Read: Privacy notice.
Loading comments…