Z-Image Turbo NSFW v2 — GGUF (Q8_0)
This is a direct GGUF conversion of tewea/z_image_turbo_bf16_nsfw (z_image_turbo_bf16_nsfw_v2.safetensors), quantized to Q8_0 to reduce VRAM usage while staying close to the original bf16 quality.
The conversion was done with city96's ComfyUI-GGUF conversion tools.
⚠️ Content notice: the base model is an NSFW finetune of Z-Image Turbo. Not for all audiences. Intended for adults only — you are responsible for complying with your local laws and the platform's terms of use.
Files
| File | Quant | Size |
|---|---|---|
z_image_nsfw_v2-Q8_0.gguf |
Q8_0 | ~6.8 GB |
Usage (ComfyUI)
- Install the ComfyUI-GGUF custom node.
- Place the
.gguffile inComfyUI/models/unet/(ormodels/diffusion_models/). - Load it with the Unet Loader (GGUF) node in place of the regular diffusion model loader.
- Use the same text encoder and VAE as the original Z-Image Turbo model.
As a Turbo-distilled model, it works with low step counts (~8 steps) and low CFG.
Credits & license
- Base finetune: tewea/z_image_turbo_bf16_nsfw (Apache 2.0)
- Original model: Tongyi-MAI/Z-Image-Turbo (Apache 2.0)
- GGUF conversion tooling: city96/ComfyUI-GGUF
Redistributed under Apache 2.0, same as the base model. All credit for the finetune goes to the original author; this repo only provides a quantized conversion.