# Qwen-Image-2.1 - Testing and Experience — retained record

> Retained reddit record from reddit:comfyui.

Canonical: https://highsignal.app/data/records/657917e1f22dc8b1

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:comfyui)  published 2026-09-25 13:54 ingested 2026-09-26 02:38

[original source ↗](https://www.reddit.com/r/comfyui/comments/1wpwyri/qwenimage21_testing_and_experience/)

## retained text



```
This is a post where I show you my experiences with the new qwen model. I'm testing a lot and if I find something that works well I'll post it here. Of course, you are also welcome to post your experiences.

First, I made a cfg overview so that you can see the behavior of cfg 1 - cfg 5.

What I found out is that the negative prompts work very well and you get significantly better image quality. My personal favorite is cfg3.

I will post a lot here about how to improve the model and am happy to answer your questions:) 
```



document fetched 2026-09-26 00:41 · canonical https://reddit.com/r/comfyui/comments/1wpwyri/qwenimage21_testing_and_experience
