Z-Image Turbo VRAM Requirements
Z-Image Turbo needs about 6.8 GB of VRAM with FP8 weights at 1024×1024, loading one model at a time. In FP16 it needs 12.5 GB (23.0 GB with everything loaded), and GGUF Q4 brings it to 4.9 GB. The smallest consumer GPU that runs it is the RTX 3050 8GB with Q4 weights, or the RTX 3050 8GB in FP8.
6.8 GB peak VRAM
Z-Image Turbo at 1024×1024 with FP8 weights, loading one model at a time (ComfyUI default on smaller cards). Keeping everything loaded needs 13.6 GB.
- Diffusion model 5.7 GB
- Text encoders 3.7 GB
- VAE 0.2 GB
- Peak activations 3.1 GB
- Overhead 0.8 GB
GPU fit
- RTX 4060 8GBFitswith swapping
- RTX 3060 12GBFitswith swapping
- RTX 4060 Ti 16GBFitsall loaded
- RTX 4090 24GBFitsall loaded
- RTX 5090 32GBFitsall loaded
- RTX 6000 Ada 48GBFitsall loaded
- RTX PRO 6000 Blackwell 96GBFitsall loaded
- Apple M4 Max 128GBFitsall loaded
The VAE decode is the peak here. Turn on tiled VAE decoding to cut it to about 0.8 GB.
Z-Image Turbo VRAM by precision
At 1024×1024. Text encoders in FP8 except for the FP16 row.
| Weights | Diffusion model | Peak (swapping) | All loaded | Smallest tier |
|---|---|---|---|---|
| FP16 / BF16 | 11.5 GB | 12.5 GB | 23.0 GB | 16 GB |
| FP8 | 5.7 GB | 6.8 GB | 13.6 GB | 8 GB |
| GGUF Q8_0 | 6.1 GB | 7.1 GB | 13.9 GB | 8 GB |
| GGUF Q5_K_S | 4.0 GB | 5.0 GB | 11.8 GB | 6 GB |
| GGUF Q4_K_S / NF4 | 3.3 GB | 4.9 GB | 11.1 GB | 6 GB |
VRAM by resolution
FP8 weights, with model swapping.
| Resolution | Latent tokens | Activations | Peak VRAM |
|---|---|---|---|
| 1024×1024 | 4,096 | 0.2 GB | 6.8 GB |
| 1536×1536 | 9,216 | 0.5 GB | 8.0 GB |
About Z-Image Turbo
A 6B distilled model that makes photorealistic images in 8 steps without classifier-free guidance. It became popular because it gives high quality on 8–12 GB cards and runs fast. The text encoder is Qwen3 4B, which can be loaded in FP8 or GGUF to save memory.
| Component | Size |
|---|---|
| Diffusion model | 6.15B parameters (Single-stream diffusion transformer) |
| Text encoders | Qwen3 4B (4B) |
| VAE | 0.084B parameters |
| Classifier-free guidance | Distilled (1 pass per step) |
| Released | 2025-11 |
| License | Apache 2.0 |
Z-Image Turbo VRAM: common questions
Can Z-Image Turbo run on an 8GB GPU?
Yes with FP8 or GGUF weights and model swapping. The 6B transformer is about 6 GB in FP8 and about 3.5 GB in GGUF Q4.
How much VRAM does Z-Image Turbo need?
At its default 1024×1024: about 12.5 GB in FP16 with model swapping (23.0 GB with everything loaded), 6.8 GB in FP8 and 4.9 GB with GGUF Q4 weights.