This section provides comprehensive guidance on how to add a new model to vLLM-Omni.
- Adding a New Model Guide: Complete step-by-step guide using Qwen3-Omni as an example
The guide covers:
- Directory structure and organization
- Implementing stage components (thinker, talker, code2wav)
- Creating the unified model class
- Model registration
- Stage configuration
- Stage input processors
- Testing strategies
For a quick reference, see the Adding a New Model Guide which walks through the complete implementation of Qwen3-Omni, a multi-stage omni-modality model.