flutter_gemma_mediapipe 1.0.1

MediaPipe (.task) on-device inference engine for flutter_gemma (Android/iOS/Web). Opt-in package; provides an InferenceEngineProvider. Bundles Google's MediaPipe Pod/Gradle deps.

example/README.md

flutter_gemma_mediapipe example

flutter_gemma_mediapipe is an opt-in inference engine for flutter_gemma. It runs MediaPipe .task / .bin models on Android, iOS, and Web. Register the engine once at startup, then use the unchanged inference API.

import 'package:flutter/widgets.dart';
import 'package:flutter_gemma/flutter_gemma.dart';
import 'package:flutter_gemma_mediapipe/flutter_gemma_mediapipe.dart';

Future<void> main() async {
  WidgetsFlutterBinding.ensureInitialized();

  // Opt into the MediaPipe engine (handles ModelFileType.task / .bin).
  await FlutterGemma.initialize(
    inferenceEngines: [MediaPipeEngine()],
  );

  // Install a .task model (downloads + sets it active).
  await FlutterGemma.installModel(
    modelType: ModelType.gemmaIt,
    fileType: ModelFileType.task,
  ).fromNetwork('https://example.com/gemma3-1b-it.task').install();

  // Create a model + session and generate.
  final model = await FlutterGemma.getActiveModel(maxTokens: 1024);
  final session = await model.createSession();
  await session.addQueryChunk(const Message(text: 'Hello!', isUser: true));
  final reply = await session.getResponse();
  print(reply);

  await session.close();
  await model.close();
}

Pass MediaPipeEngine() alongside other engines (e.g. LiteRtLmEngine from flutter_gemma_litertlm) if your app uses both .task and .litertlm models. On web the MediaPipe runtime loads from a CDN — see the package README for the web/index.html setup. A full runnable app lives in the flutter_gemma example.