foundation-models-on-device

On-device LLM integration for iOS 18+ using Apple's FoundationModels framework with privacy-first text generation and structured output. Patterns for integrating Apple's on-device language model into apps using the FoundationModels framework. Covers text generation, structured output with @Generable , custom tool calling, and snapshot streaming — all running on-device for privacy and offline support. Always check model availability before creating a session: Key points for instructions: Generate structured Swift types instead of raw strings: Let the model invoke custom code for domain-specific tasks: Stream structured responses for real-time UI with PartiallyGenerated types: - Covers text generation, structured output via @Generable macro, custom tool calling, and snapshot streaming—all running locally without cloud dependency - Requires availability checks before session creation; supports single-turn and multi-turn conversations with optional system instructions - Guided generation with @Guide constraints (numeric ranges, array counts, semantic hints) produces strongly-typed Swift structs instead of raw strings - Tool calling lets models invoke custom functions for domain-specific tasks; streaming mode progressively yields partial results for real-time UI updates - 4,096 token context limit applies across instructions, prompt, and output combined; single request per session...