# Run Stable Diffusion XL

Turns a written description into a finished image. Ready in about 90 seconds on an RTX 4090, from $0.45/hr. No install, no code.

## What it does

Stable Diffusion XL makes images from text. You describe what you want and it draws it, at roughly the size and quality of a stock photo.

It is the most widely used open image model, which means it understands a very broad range of subjects and styles. It also has the largest library of add-ons, so you can teach it a specific product, face, or art style later.

It is the safe first choice for product shots, marketing images, concept art, and social posts.

## Your setup

We run Stable Diffusion XL on an RTX 4090 with 24GB of memory, at $0.45 per hour, billed by the minute. It takes about 8 seconds per image.

Stable Diffusion XL needs about 12GB of memory to run at full size without swapping parts of itself in and out. The RTX 4090 has 24GB, so the model and its refiner both stay resident and every image runs at full speed. Cheaper 8GB cards can run it, but they offload to system memory and each image takes three to four times longer, which costs more overall despite the lower hourly rate.

## What it costs

500 images takes about 67 minutes of GPU time, which is about $0.50.

## Two ways to run it

- **Open the app.** Open ComfyUI in your browser with a working image graph already loaded. Type a prompt, press go. Best for exploring and dialing in a look.
- **Run a batch job.** Upload a spreadsheet of prompts. We generate every image, hand back a zip, and shut the GPU off. Best for producing a catalog in one pass.

## Model facts

| Field | Value |
|---|---|
| Parameters | 3.5B (base UNet) |
| License | CreativeML Open RAIL++-M |
| Memory required | 12GB minimum, 24GB recommended |
| Base model | Trained from scratch by Stability AI |
| Publisher | Stability AI |
| Recommended hardware | RTX 4090, 24GB |
| Category | [Image generation](https://gpuvault.io/working-on/image-generation/) |

## Questions people ask

### Can I sell the images Stable Diffusion XL makes?

Yes. Stable Diffusion XL is released under the CreativeML Open RAIL++-M license, which permits commercial use of the images you generate. The license restricts what you may use the model for, not what you may do with the output. You are still responsible for not generating someone else’s trademark or a real person’s likeness and selling it.

### Do I need to write prompts in a special way?

No. A plain sentence works. "A ceramic coffee mug on a linen tablecloth in morning light" is a complete prompt. Adding words like "studio photo" or "watercolor" steers the style. The workspace we hand you has a starting prompt already in it so you can edit rather than start from a blank box.

### Do I need to install anything to use Stable Diffusion XL?

No. We start a machine with Stable Diffusion XL already loaded and hand you a link. Everything runs in your browser, there is nothing to download, and nothing is left on your computer afterwards. It works the same on a Mac, a Windows laptop, or a Chromebook. A workspace is usually ready in about 90 seconds.

### How much does it cost to run 500 images?

About $0.50. Stable Diffusion XL takes about 8 seconds per image on the RTX 4090 we recommend, so 500 images is roughly 67 minutes of GPU time at $0.45 per hour. Billing is by the minute, and a batch job shuts the GPU off the moment the last item finishes. The calculator above works this out for your own numbers.

### Can I run Stable Diffusion XL on a cheaper card?

Sometimes, and the calculator will not always make it look worth it. Stable Diffusion XL needs about 12GB of memory to run at full size without swapping parts of itself in and out. The RTX 4090 has 24GB, so the model and its refiner both stay resident and every image runs at full speed. Cheaper 8GB cards can run it, but they offload to system memory and each image takes three to four times longer, which costs more overall despite the lower hourly rate. If you want to try a different card anyway, the advanced catalog lets you pick one and shows the estimated time before you commit.

### What are the license restrictions on Stable Diffusion XL?

Stable Diffusion XL is released under the CreativeML Open RAIL++-M. That is not a permissive license, so read it before you use the output in paid work. We show the license on every model page precisely because this catches people out, and there is usually a permissively licensed alternative in the same category.

---

*Source: https://gpuvault.io/model/stable-diffusion-xl/ — GPUVault, GPU rental by the minute.*