A Flutter plugin enabling on-device text embedding generation using a lightweight, locally run model inspired by EmbeddingGemma. Designed for semantic search and similarity tasks without relying on external APIs. The package runs entirely on Android, iOS, and Web platforms, ensuring privacy and low latency. It allows developers to process and compare text embeddings directly within the app, ideal for applications requiring real-time, offline-capable AI features.
Use cases
- Offline semantic search
- Text similarity comparison
- Private data processing
- Local AI inference
- Content recommendation systems
Key features
- On-device inference
- Lightweight model integration
- Cross-platform support (Android, iOS, Web)
- Privacy-preserving
- Low-latency embeddings
Suitable for
- Apps needing offline AI capabilities
- Developers prioritizing user privacy
- Semantic search implementations
- Mobile-first generative AI projects
- Educational or prototyping tools
Considerations
- Higher memory usage on mobile devices
- Limited model size compared to cloud models
- Requires device with sufficient computational resources
- Initial download size of model assets
- Performance varies across older hardware