# Qwen-Image requirements

Updated: 2026-06-15

20B MMDIT image model, released 2025-08.

- Peak VRAM: ~14 GB at Q4_K_M GGUF (~57 GB all-resident)
- Offload floor: ~3 GB, much slower
- Resolution: 1328×1328
- Tools: ComfyUI, Nunchaku (SVDQuant 4-bit)
- License: Apache-2.0 (commercial use: yes)
- Runs on 15 of 39 tracked devices

Sources: https://huggingface.co/Qwen/Qwen-Image, https://huggingface.co/city96/Qwen-Image-gguf, https://docs.comfy.org/tutorials/image/qwen/qwen-image, https://sandner.art/qwen-image-and-edit-local-gguf-generations-with-lightning/, https://github.com/QwenLM/Qwen-Image
More: https://localmodel.run/model/qwen-image