# Stable Diffusion 3 Medium - Documentation

> For the complete documentation index, see [llms.txt](/llms.txt). Markdown is available with `Accept: text/markdown` and `.md` URL variants.

Source: /docs/guides/model-apis-stable-diffusion-3-medium

# Stable Diffusion 3 Medium

Stability AI recently released the weights for Stable Diffusion 3 Medium, a 2 billion parameter text-to-image model that excels at photorealism, typography, and prompt adherence.
Now you can run the official Stable Diffusion 3 Medium model on Novita AI.
You can try this model at the [Playground](https://novita.ai/model-api/product/sd3), or integrate it into your product using the [Text to Image API](/docs/api-reference/model-apis-txt2img) and [Image to Image API](/docs/api-reference/model-apis-img2img).

##

[​](#api-request-example)

API Request Example

```
curl --location 'https://api.novita.ai/v3/async/txt2img' \
--header 'Authorization: Bearer {{API Key}}' \
--header 'Content-Type: application/json' \
--data '{
"extra": {
"response_image_type": "jpeg"
},
"request": {
"model_name": "sd3_base_medium.safetensors",
"prompt": "A heart logo with the text \"Novita AI\" on it",
"width": 1024,
"height": 1024,
"image_num": 1,
"steps": 28,
"seed": -1,
"clip_skip": 1,
"guidance_scale": 4.5,
"sampler_name": "FlowMatchEuler"
}
}'
```

##

[​](#please-note)

Please Note

- Currently, only the checkpoint `sd3_base_medium.safetensors` is available for Stable Diffusion 3, and no LoRAs or ControlNets are supported.

- You can only use the sampler `FlowMatchEuler` for Stable Diffusion 3.

##

[​](#best-practices-for-stable-diffusion-3)

Best Practices for Stable Diffusion 3

- The longer and more complex the prompt, the more likely something will be missing.

- Please avoid using negative prompts, as they are ineffective.

- Suggested parameters:

Steps: Use around 28 steps.

- Guidance Scale: Set between 3.5 and 4.5 for balanced adherence and image quality.

- Resolution: Use resolutions around 1 megapixel, ensuring they are divisible by 64.

Last modified on January 14, 2026