React-Friendly API
Simple hooks for loading and running models. No complex setup required. Just wrap your app with LiteRtProvider and start using models.
Automatic Acceleration
Automatically selects the best accelerator available. WebGPU for maximum performance, with seamless fallback to WASM when needed.
TensorFlow.js Integration
Works seamlessly with TensorFlow.js tensors. Input and output are tf.Tensors, making it easy to integrate with existing ML workflows.