stability ai

SD Turbo

Stable

by Stability AI

Adversarial distillation of SD 1.5 for single-step image generation. Only 0.86B UNet — the smallest and fastest Stable Diffusion variant. Quality is lower than SD 1.5 but generation is nearly instant. Ideal for real-time interactive use.

VRAM requirements, GPU fit, and setup notes for SD Turbo, including 8GB/12GB fit guidance where relevant. Recommended runtimes: ComfyUI and Diffusers support. Best download size: ~2.5 GB at FP16.

  • Single-step generation — near instant
  • Only 0.86B params — runs on almost any GPU
  • Based on SD 1.5 architecture
  • Quality tradeoff for maximum speed
ComfyUI, DiffusersFP16 safetensors

Your hardware

Detecting...

Parameters0.86B
Max Resolution512×512
Default Steps1
ArchitectureUNET
Licensestability-ai-non-commercial

Image Quality Benchmarks

Measured quality metrics for SD Turbo outputs.

Human Preference Score55%

How often humans prefer this model's output (0-100%)

Aesthetic Score5.5

Visual quality and composition rating (5-9 scale)

VRAM Requirements by Resolution and Precision

Compare which GPUs can run SD Turbo at different precisions. FP8 uses less memory than FP16 when available, and the grade shows how comfortably each GPU handles the workload.

FP16 (full precision)

ResolutionVRAM RequiredRTX 4090 24GBRTX 3060 12GBRTX 4060 8GBMacBook Pro M4 Pro 24GB
512×5123.5 GBSSSS
768×7684.2 GBSSSS
1024×10245.2 GBSSSS

Optimization Tips

ControlNets available

Add guided generation with 3 adapters (+0.7 GB VRAM each)

Rich LoRA ecosystem

Customize style, characters, and quality with community LoRAs

Run with Python

Run with Python (diffusers)
from diffusers import StableDiffusionPipeline
import torch

pipe = StableDiffusionPipeline.from_pretrained(
    "stabilityai/sd-turbo",
    torch_dtype=torch.float16
)
pipe.to("cuda")

image = pipe(
    prompt="your prompt here",
    num_inference_steps=1,
    height=512,
    width=512,
).images[0]
image.save("output.png")

Get started

Setup instructions for running SD Turbo locally

1. Download the model

Get the checkpoint from HuggingFace

2. Place in:

ComfyUI/models/checkpoints/

3. Launch ComfyUI

python main.py

Memory Breakdown

VRAM allocation at 1024×1024 on RTX 4090 24GB (24 GB)

Required: 5.2 GBAvailable: 24.0 GB
Weights1.7 GB
VAE0.2 GB
Text Encoder0.2 GB
Activations2.0 GB
Overhead0.5 GB

Estimated Generation Time

Time per image at 1024×1024, 28 steps, FP16.

RTX 4090 24GB500ms
RTX 3060 12GB~1.9s
RTX 4060 8GB~2.8s
MacBook Pro M4 Pro 24GB~4s

Sample Outputs

Available Formats, Downloads & Setup

Download SD Turbo in the precision that matches your GPU. Lower precision usually means less VRAM pressure, while higher precision keeps more quality.

フォーマット精度サイズプロバイダー
safetensors推奨FP162.5 GBofficialダウンロード

ControlNet Support

3 ControlNets available for SD Turbo. ControlNets add guided image generation (edges, depth, pose) at the cost of extra VRAM.

Canny Edge

+0.7 GB VRAM

Edge-based structural guidance. SD 1.5 ControlNet compatible.

comfyuiautomatic1111diffusers
View on HF

Depth Map

+0.7 GB VRAM

Depth-based 3D spatial control.

comfyuiautomatic1111diffusers
View on HF

OpenPose

+0.7 GB VRAM

Human pose estimation for character control.

comfyuiautomatic1111diffusers
View on HF

LoRA Ecosystem

Massive Ecosystem

Full SD 1.5 LoRA compatibility

Fine-tune of sd-1-5

Related Workflows

You might also like

Frequently asked questions

FAQ — SD Turbo VRAM, Runtimes & Fit

How much VRAM does SD Turbo need?

SD Turbo (0.86B parameters) requires approximately 5.2 GB of VRAM at FP16 precision for standard 1024×1024 image generation. If you want a lighter setup, lower precisions like FP8 can reduce memory use when available.

Can I run SD Turbo on an 8GB GPU?

SD Turbo usually needs more than 8GB for comfortable local use. Check the VRAM table above for the exact resolution and precision trade-off.

Does SD Turbo work in ComfyUI and Diffusers?

SD Turbo is marked for ComfyUI and Diffusers support in our catalog, so those are the runtimes we recommend first for local setup. If your workflow uses another front end, check the model's available formats and workflow notes above before downloading.

Can I run SD Turbo on RTX 4090?

Yes, the RTX 4090 (24 GB VRAM) can run SD Turbo comfortably at FP16. Expected generation time is around 500ms per image at 1024×1024.

Does SD Turbo support ControlNet?

Yes, SD Turbo has 3 ControlNet adapters available: Canny Edge, Depth Map, OpenPose. Each ControlNet adds roughly 0.7 GB of extra VRAM.

Does SD Turbo have LoRA support?

Full SD 1.5 LoRA compatibility The LoRA ecosystem for SD Turbo is rated as "massive". Each LoRA adds roughly 0.1 GB of extra VRAM.

How fast is SD Turbo?

On a reference GPU (RTX 4090 24GB), SD Turbo generates a 1024×1024 image in approximately 500ms at FP16 with 28 inference steps. Faster GPUs with higher memory bandwidth will produce images more quickly.

About SD Turbo

Use cases
fast-generationreal-timeprototypinglightweight
Recommended runtimes
comfyuidiffusers

See also