Pruna Qwen-Image 2.1 (8 steps) is now available in ZPix
retained text
As a single model for your convenience. On my RTX 3070M 8GB, once warmed, this distilled model takes 20s to generate a 1024x1024 image. Base model is available as well. LoRAs are supported on both (base model supports them better). Download at: [https://github.com/SamuelTallet/ZPix](https://github.com/SamuelTallet/ZPix)
document fetched 2026-09-27 00:31 · canonical https://reddit.com/r/StableDiffusion/comments/1wr440n/pruna_qwenimage_21_8_steps_is_now_available_in