Skip to content
Sign in
  • Browse all toolsFilter by category and pricing
  • Browse categoriesWriting, code, image, audio and more
  • Compare toolsPut up to three side by side
  • Read the blogGuides and comparisons

SDXL Turbo

Real-time text-to-image generation in a single denoising step

SDXL Turbo is a text-to-image model that generates images from prompts in real time using a single denoising step. It is available to researchers and enthusiasts through a free Clipdrop beta, with model weights and code on Hugging Face under a license that permits personal, non-commercial use.

Updated 22 Nov 20267 sections

At a glance

Pricing
FreeThe Clipdrop beta is free to try. Downloadable weights and code are licensed for personal, non-commercial use. Contact Stability AI about commercial use.
Added
22 Nov 2026
SDXL Turbo website screenshot

What is SDXL Turbo?

SDXL Turbo is a real-time text-to-image model from Stability AI that turns text prompts into images. Built on SDXL 1.0, SDXL Turbo uses Adversarial Diffusion Distillation to generate an image in a single denoising step. Stability AI provides a browser demonstration for people who want to try the model and downloadable files for researchers and enthusiasts who want to examine or experiment with the release.

What SDXL Turbo can do

SDXL Turbo reduces the number of denoising steps needed to produce an image from a prompt. Stability AI reported generating a 512x512 image in 207ms on an A100 using fp16. That timing covers the full generation process in the reported test: encoding the prompt, performing a single denoising step, and decoding the image.

  • SDXL Turbo generates images from text prompts in real time.
  • SDXL Turbo can produce an image in a single denoising step.
  • SDXL Turbo can generate 512x512 images, as shown in Stability AI’s benchmark.

How SDXL Turbo works

SDXL Turbo combines adversarial training and score distillation through a technique called Adversarial Diffusion Distillation. Stability AI describes the technique as a way to produce single-step outputs while avoiding artifacts or blurriness that can occur with other distillation methods. The research paper covers the technique in more detail, while the Clipdrop beta demonstrates real-time generation in most browsers.

Who SDXL Turbo is for

SDXL Turbo is aimed at researchers and enthusiasts interested in fast text-to-image generation. The Clipdrop beta is an option for people who want to see how the model responds to text prompts in a browser. The Hugging Face release provides weights and code for people who want to work with the model files, and the research paper covers the distillation method.

  • Generate images from text prompts in Clipdrop.
  • Experiment with the weights and code for non-commercial research.
  • Study the distillation technique in the research paper.

SDXL Turbo pricing and plans

SDXL Turbo is free to try through the Clipdrop beta. Stability AI lists the model weights and code for download on Hugging Face under a non-commercial research license that permits personal, non-commercial use. The release page does not list a price or plan for commercial use and instead directs prospective commercial users to contact Stability AI.

Strengths and limitations

SDXL Turbo outperformed a 4-step LCM-XL configuration in Stability AI’s reported blind tests using a single step. A 4-step SDXL Turbo configuration also outperformed a 50-step SDXL configuration. Evaluators compared outputs generated with the same prompt, judging prompt adherence separately from image quality. The speed figure on the release page was measured on an A100 and applies to that test setup.

  • The published research license limits use to personal, non-commercial purposes.
  • The Clipdrop demonstration is a beta.
  • The reported generation time is specific to the benchmark setup.

Getting started with SDXL Turbo

SDXL Turbo can be tried through the Clipdrop beta by entering a text prompt in the browser demonstration. For work with the model files, Stability AI links to the weights and code on Hugging Face under the non-commercial research license. The release page also links to the research paper for readers who want to understand the training method before experimenting with the model.

SDXL Turbo FAQ

What is SDXL Turbo?

SDXL Turbo is a real-time text-to-image model from Stability AI that can generate an image from a text prompt in a single denoising step. SDXL Turbo builds on SDXL 1.0 and uses a technique called Adversarial Diffusion Distillation.

Is SDXL Turbo free?

SDXL Turbo is free to try through Stability AI’s Clipdrop beta. The weights and code are available on Hugging Face under a research license that permits personal, non-commercial use. Stability AI asks prospective commercial users to contact the company.

Who is SDXL Turbo for?

SDXL Turbo is for researchers and enthusiasts interested in real-time text-to-image generation. Researchers can read about the distillation method or download the weights and code for non-commercial use, while enthusiasts can try the model in the Clipdrop browser demonstration.

How fast is SDXL Turbo?

SDXL Turbo generated a 512x512 image in 207ms on an A100 in Stability AI’s reported benchmark. The measurement used fp16 and included prompt encoding, a single denoising step, and image decoding.

Where can I download SDXL Turbo?

SDXL Turbo model weights and code are available to download from Hugging Face under a research license that permits personal, non-commercial use. Stability AI also offers a Clipdrop beta for trying the model in a browser without downloading the weights or code.

Can I use SDXL Turbo commercially?

SDXL Turbo is not licensed for commercial use under the non-commercial research license described on the release page. Stability AI directs people who want to use the model for commercial products or purposes to contact the company.

From the community

No reviews yet

Be the first to review this tool. What you write helps the next person choose.

More tools like SDXL Turbo

Zizoto

www.zizoto.com

Generate AI images with SDXL, remix community creations, and print posters.

Free

Peacasso

www.victordibia.com

Peacasso provides a prompt-based interface for generating AI art and experimenting with diffusion models.

Freemium

SDXL emoji

www.replicate.com

Generate emoji-style images from text prompts with an SDXL model fine-tuned on Apple emojis.

Paid

Stablecog

www.stablecog.com

Generate AI art from text descriptions or input images, with free and paid plans.

Freemium