What is the @Generable macro used for?

The @Generable macro is used to define structured Swift types that the AI can populate directly, providing compile-time safety and eliminating the need for manual string parsing.

What is the token limit for Apple's on-device models?

The models generally have a 4,096-token context window which includes the system instructions, the user prompt, and the generated response.

Which iOS version supports the FoundationModels framework described here?

These patterns and capabilities are designed for iOS 26 and later, which introduced enhanced support for Apple Intelligence and on-device generative models.

What are the benefits of on-device LLMs over cloud-based APIs?

On-device LLMs provide superior privacy because data never leaves the device, they support offline functionality, and they eliminate the latency and costs associated with cloud API calls.

How does snapshot streaming differ from incremental streaming?

Snapshot streaming provides a complete 'partially generated' version of the final object with every update, making it easier to update complex UI components compared to handling individual text chunks.

Apple FoundationModels On-Device

Name: Apple FoundationModels On-Device
Author: Fabio29T

byFabio29T

0•

Mobile Development

Integrates Apple's on-device language models into iOS applications for privacy-preserving text generation, structured output, and tool calling.

This skill provides specialized guidance for implementing Apple's FoundationModels framework in iOS 26+ environments. It enables developers to build high-performance, privacy-first AI features that run entirely on-device, supporting offline usage and secure data handling. By leveraging core patterns like @Generable for structured data, custom tool calling for domain-specific logic, and snapshot streaming for real-time UI updates, developers can seamlessly integrate Apple Intelligence into their mobile workflows while managing hardware constraints and model availability.

Características Principales

010 GitHub stars

02Device-side model availability and eligibility checking patterns

03Structured data extraction using @Generable and @Guide constraints

04Custom tool calling implementation for domain-specific AI tasks

05Multi-turn conversation management with persistent session context

06Real-time snapshot streaming for fluid UI updates

Casos de Uso

01Building privacy-focused AI assistants that process sensitive user data locally

02Implementing offline-capable AI features for text summarization and content generation

03Extracting structured objects like calendar events or contacts from natural language

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add fabio29t/everything-claude foundation-models-on-device

For use in Claude.ai and ChatGPT

Características Principales

010 GitHub stars

02Device-side model availability and eligibility checking patterns

03Structured data extraction using @Generable and @Guide constraints

04Custom tool calling implementation for domain-specific AI tasks

05Multi-turn conversation management with persistent session context

06Real-time snapshot streaming for fluid UI updates

Casos de Uso

01Building privacy-focused AI assistants that process sensitive user data locally

02Implementing offline-capable AI features for text summarization and content generation

03Extracting structured objects like calendar events or contacts from natural language

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add fabio29t/everything-claude foundation-models-on-device

For use in Claude.ai and ChatGPT