Flutter_Gemma is a Flutter plugin enabling local execution of Gemma and other large language models (LLMs) across Android, iOS, Web, and desktop platforms. It supports advanced features like multimodal input (vision and audio), function calling, thinking mode, GPU acceleration, text embeddings, and on-device retrieval-augmented generation (RAG). Designed for privacy-focused applications, it runs models entirely on the device without relying on cloud services. Ideal for developers building AI-powered apps with real-time, secure, and offline capabilities.
Use cases
- Offline AI chat applications
- Privacy-first content generation
- Multimodal image and audio processing
- On-device RAG systems
- Real-time function calling in apps
- Local LLM inference for mobile and desktop
Key features
- Runs LLMs locally on all platforms
- Supports vision and audio multimodal input
- GPU acceleration for performance
- Function calling and thinking mode
- Text embeddings and on-device RAG
- Secure, privacy-preserving inference
Suitable for
- Developers building private AI apps
- Teams needing offline LLM capabilities
- Apps requiring low-latency inference
- Projects focused on data privacy
- Cross-platform AI solutions
Considerations
- High memory usage for large models
- Requires compatible hardware for optimal performance
- Model size impacts app bundle size
- Limited to supported model architectures
- Initial setup may require platform-specific configurations