11 models · updated 2026-09-02
Text to image models: what your GPU can actually run
A diffusion pipeline is several networks, not one, and the memory a local text to image model needs depends as much on how you arrange them as on how big they are.
Results
| Model | Denoiser | Verdict | Arrangement | Peak VRAM | Time |
|---|---|---|---|---|---|
| Loading… | |||||
Every model in this catalog
Download sizes are the full pipeline at the precision it is published in.
Stable Diffusion XL
2.567B denoiser · 3.469B total · 12.9 GB to download
Stable Diffusion 1.5
1.719B denoiser · 1.926B total · 7.2 GB to download
DreamShaper 7
0.86B denoiser · 1.067B total · 4.0 GB to download
SDXL Turbo
2.567B denoiser · 3.469B total · 12.9 GB to download
FLUX.1 dev
11.901B denoiser · 16.87B total · 31.4 GB to download
Z-Image Turbo
6.155B denoiser · 10.261B total · 30.6 GB to download
FLUX.1 schnell
11.891B denoiser · 16.86B total · 31.4 GB to download
RealVisXL V5
2.567B denoiser · 3.469B total · 12.9 GB to download
Qwen-Image
20.43B denoiser · 28.849B total · 53.7 GB to download
Stable Diffusion 3.5 Medium
2.47B denoiser · 8.135B total · 15.1 GB to download
Stable Diffusion 3.5 Large
8.146B denoiser · 13.811B total · 25.7 GB to download