sglang-diffusion-add-model

What problem does it solve?

This guide enables you to add a new diffusion model or Diffusers pipeline to SGLang, unlocking seamless extension of the multimodal generation system.

Core Features & Use Cases

• Hybrid-style integration: choose Hybrid or Modular approach to tailor pre-processing and denoising stages for new diffusion variants.
• End-to-end wiring: define DiT config, VAE, schedulers, and pipeline callbacks, then register with the model registry.
• Extensibility: reuse existing components when possible to minimize duplication and maximize compatibility with Diffusers.

Quick Start

Create and integrate a new diffusion model into SGLang's multimodal_gen framework.