From the source
Amazon published a tutorial showing how to deploy the Qwen3-TTS model on SageMaker AI using the vLLM-Omni Deep Learning Container.
- The setup streams text input and audio output over a persistent bidirectional connection, enabling speech playback before the full response is generated.
- The walkthrough uses a Gradio application and supports instance pooling with fallback types such as ml.g6.xlarge, ml.g6e.xlarge, ml.g5.xlarge, or ml.g4dn.xlarge.





