Stability.ai¶

1. Overview¶

Stability AI develops AI models, and its best-known family is Stable Diffusion, a deep-learning image generation model widely used for creating high-quality images from text descriptions.

Description：

Models in this family are not optimized for Chinese prompts. Use English natural-language prompts when possible.

2. Request¶

Method:POST

Endpoint:

https://gateway.serevixai.ai/v1/images/generations

3. Parameters¶

3.1 Header Parameters¶

Parameter	Type	Required	Description	Example
`Content-Type`	string	Yes	Sets the request content type. It must be `multipart/form-data`.	`multipart/form-data; boundary=------------------------y0QNRDhXQUE3yHQi7DEbLz`
`Accept`	string	Yes	Sets the response content type. It must be `image/*`.	`image/*`
`Authorization`	string	Yes	API key required for authentication, in the format `Bearer $YOUR_API_KEY`.	`Bearer $YOUR_API_KEY`

3.2 Body Parameters (multipart/form-data)¶

Parameter	Type	Required	Description	Example (default)
model	string	Yes	The model ID to use. See Model List for available versions, such as `sd3-large`.	`sd3-large`
prompt	string	Yes	A text prompt describing the image to generate. Maximum prompt length is 10,000 characters.	`A cute baby sea otter`
mode	string	No	Controls whether the request runs in `text-to-image` or `image-to-image` mode. `text-to-image` requires only `prompt`, while `image-to-image` requires `prompt`, `image`, and `strength`. `sd-core` does not support `image-to-image`.	`text-to-image`
image	string(binary)	No	Input image used as the starting point for `image-to-image`. Supported formats: `jpeg`, `png`, and `webp`. Each side must be at least 64 pixels.
strength	number[0,1]	No	Sometimes called denoising, this controls how much the source image influences the result. A value of `0` keeps the result identical to the input image, while `1` behaves as if no image was supplied.	`0.9`
negative_prompt	string	No	Keywords describing content you do not want to appear in the output image. `sd3-large-turbo` does not support this parameter.	`a cat`
seed	number[0,4294967294]	No	Specific seed value used to control randomness. If omitted or set to `0`, a random seed is used.	`0`
aspect_ratio	string	No	Controls the generated image aspect ratio. Supported values: `16:9`, `1:1`, `21:9`, `2:3`, `3:2`, `4:5`, `5:4`, `9:16`, and `9:21`.	`1:1`
output_format	string	No	Output image format. Supported values are `jpeg` and `png`.	`png`
style_preset	string	No	Guides the image model toward a specific style. Supported values include `3d-model`, `analog-film`, `anime`, `cinematic`, `comic-book`, `digital-art`, `enhance`, `fantasy-art`, `isometric`, `line-art`, `low-poly`, `modeling-compound`, `neon-punk`, `origami`, `photographic`, `pixel-art`, and `tile-texture`. Only `sd-core` supports this parameter.

4. Request Examples¶

text-to-image

curl https://gateway.serevixai.ai/v1/images/generations \
    -H "Accept: image/*" \
    -H "Authorization: Bearer $YOUR_API_KEY" \
    -F model="sd3-large" \
    -F prompt="A cute baby sea otter wearing a beret"

image-to-image

curl https://gateway.serevixai.ai/v1/images/generations \
    -H "Accept: image/*" \
    -H "Authorization: Bearer $YOUR_API_KEY" \
    -F model="sd3-large" \
    -F prompt="A cute baby sea otter wearing a beret" \
    -F mode="image-to-image" \
    -F image="@input.png" \
    -F strength=0.9 \

5. Response Example¶

HTTP/1.1 200 OK
Content-Type: image/png

Binary image file