flutter_edge_ai_builtin_ai 0.3.2

Opt-in flutter_edge_ai engine for the OS built-in model: Gemini Nano (Android), Apple Foundation Models (iOS/macOS), Windows AI Foundry, Chrome's Prompt API on Web.

0.3.2

• Breaking: createChat(isThinking:) is enableThinking; require flutter_edge_ai ^2.1.0.

0.3.1

• Require flutter_edge_ai 2.x.

0.3.0

• Renamed from flutter_gemma_builtin_ai.
• Breaking: a pure-Dart adapter over flutter_local_ai ^0.2.1; the BuiltInAi* API is unchanged.
• Breaking: no longer a plugin; re-lock Podfile.lock, and macOS apps need a 12.0 target.
• Breaking: requires Flutter 3.44 / Dart 3.12; pigeon.g.dart is gone.
• Windows support through Windows AI Foundry (Phi Silica).
• supportImage: true where there is no vision fails at model creation.
• localAiModel / localAiSession expose native tools and structured output (@experimental).

0.2.2

• Windows/Linux: availability() reports unavailableDeviceUnsupported instead of throwing.

0.2.1

• Add BuiltInAiHuggingFaceResolver (auto-registered from BuiltInAiEngine) so resolveHuggingFace reports that built-in OS models have no Hugging Face file (#454).
• fix: iOS/macOS builds failed on Xcode below 26.4 — SystemLanguageModel.tokenCount is absent from those SDKs and #available cannot gate a missing declaration (#460).

0.2.0

• Web support: Gemini Nano via the Chrome Prompt API (desktop Chrome/Edge).

0.1.1

• Android plugin no longer applies KGP — Flutter's Gradle plugin does (#440).
• iOS floor lowered to 15.0 — every Foundation Models call is #available-guarded (#441).

0.1.0

• Initial release: Gemini Nano (Android) + Apple Foundation Models (iOS/macOS) engine