Foundation Models
Apple's on-device AI framework providing access to a 3B parameter language model for summarization, extraction, classification, and content generation. Runs entirely on-device with no network required.
Overview
Foundation Models enable intelligent text processing directly on device without server round-trips, user data sharing, or network dependencies. The core principle: leverage on-device AI for specific, contained tasks (not for general knowledge).
Reference Loading Guide
ALWAYS load reference files if there is even a small chance the content may be required. It's better to have the context than to miss a pattern or make a mistake.
| Reference | Load When |
|---|---|
| Getting Started | Setting up LanguageModelSession, checking availability, basic prompts |
| Structured Output | Using @Generable for type-safe responses, @Guide constraints |
| Tool Calling | Integrating external data (weather, contacts, MapKit) via Tool protocol |
| Streaming | AsyncSequence for progressive UI updates, PartiallyGenerated types |
| Troubleshooting | Context overflow, guardrails, errors, anti-patterns |
Core Workflow
- Check availability with
SystemLanguageModel.default.availability - Create
LanguageModelSessionwith optional instructions - Choose output type: plain String or @Generable struct
- Use streaming for long generations (>1 second)
- Handle errors: context overflow, guardrails, unsupported language
Model Capabilities
| Use Case | Foundation Models? | Alternative |
|---|---|---|
| Summarization | Yes | - |
| Extraction (key info) | Yes | - |
| Classification | Yes | - |
| Content tagging | Yes (built-in adapter) | - |
| World knowledge | No | ChatGPT, Claude, Gemini |
| Complex reasoning | No | Server LLMs |
Platform Requirements
- iOS 26+, macOS 26+, iPadOS 26+, visionOS 26+
- Apple Intelligence-enabled device (iPhone 15 Pro+, M1+ iPad/Mac)
- User opted into Apple Intelligence
Common Mistakes
- Using Foundation Models for world knowledge — The 3B model is trained for on-device tasks only. It won't know current events, specific facts, or "who is X". Use ChatGPT/Claude for that. Keep prompts to: summarizing user's own content, extracting info, classifying text.
- Blocking the main thread — LanguageModelSession calls must run on a background thread or async context. Blocking the main thread locks UI. Always use
Task {}or background queue. - Ignoring context overflow — The model has finite context. If the user pastes a 50KB document, it will fail silently or truncate. Check input length and trim/truncate proactively.
- Forgetting to check availability — Not all devices support Foundation Models. Check
SystemLanguageModel.default.availabilitybefore using. Graceful degradation is required. - Ignoring guardrails — The model won't answer harmful queries. Instead of fighting it, design prompts that respect safety guidelines. Rephrasing requests usually works.