ComfyUI labels Qwen as FLUX: Understanding model_type

When using Qwen models within ComfyUI, you might find yourself scratching your head over a label in the logs: model_type FLUX. You might wonder if this is correct or if it’s just noise. The answer isn’t immediately obvious because it depends on how modern loaders map architectures internally.

It turns out that model_type doesn’t describe the commercial name of a model, but rather the pipeline or architecture it’s being mapped to. In this case, Qwen Image Edit is registered as type FLUX because it uses an architecture compatible with the FLUX transformer and diffusion pipelines.

The List of Model Types

To put this in perspective, there isn’t just one or two types. ComfyUI has a stable set of classifications for these loaders. The most common ones you’ll encounter are SD1, SD2, and SDXL. However, alongside those, you have FLUX (which includes Qwen), PIXART_ALPHA, KANDINSKY, WÃœRSTCHEN, COSMOS, and even generic categories like DIT or UNKNOWN.

The Bottom Line

The key takeaway is that the label only matters for knowing which pipeline to use. If Qwen loads correctly without errors, it’s simply a matter of efficient coding—reusing existing FLUX logic rather than duplicating code. It’s not an error, and it doesn’t change what the model actually is.