import torch
from diffusers import DiffusionPipeline
# switch to "mps" for apple devices
pipe = DiffusionPipeline.from_pretrained("Ccre/Z-Image-Turbo-MXFP8", dtype=torch.bfloat16, device_map="cuda")
prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
image = pipe(prompt).images[0]Tested on NVIDIA Blackwell (RTX 5060 Ti). This MXFP8 version offers a ~1.9x speedup over BF16 with very little visual degradation.
See the full forensic benchmarks and 'Difference Maps' at: blackwell-mxfp8-nvfp4
Read the step-by-step installation guide for MXFP8 and NVFP4.
Images created at 12 steps.
- Downloads last month
- 57
Model tree for Ccre/Z-Image-Turbo-MXFP8
Base model
Tongyi-MAI/Z-Image-Turbo