Modify images using natural language instructions.
The API offers two models. We recommend v2 for most production use cases due to its superior quality and speed.
v2
Recommended. Our latest model: native 2K resolution, faster, more accurate prompt following, better texture realism and product preservation.
v1
Older model retained for workflows requiring custom aspect ratio.
v2 model is used by default.
"options": {
"model": "v2",
"prompt": "add a duck"
}The prompt is a text description that guides the AI on how to edit the image. It tells the AI what new elements to add or what changes to make.
"options": {
"prompt": "add a duck"
}The choice of inference steps is only supported for the v1 model.
Inference steps controls the number of steps the AI takes to refine the image. A higher number of steps can produce more detailed and higher-quality images but will also increase the processing time.
The choice of guidance scale is only supported for the v1 model.
The guidance scale determines how closely the AI should follow the instructions in your prompt. A higher value means the AI will stick more strictly to your prompt, while a lower value allows the AI to be more creative and generate a more diverse result.
The choice of aspect ratio is only supported for the v1 model.
The aspect ratio defines the final dimensions of the edited image. This parameter is used to crop and resize the output to a specific width-to-height ratio. The API supports various common aspect ratios.
By default, the aspect ratio is the same as an input image. But you can change it to any of the following values:
1:1
1024x1024px
2:3
832x1248px
3:2
1248x832px
3:4
880x1168px
4:3
1168x880px
9:16
768x1360px
16:9
1360x768px
9:21
656x1552px
21:9
1552x656px
model
string
"v1" , "v2"
"v2"
prompt
string
3-5000
required
inference_steps
integer
1-50
depends on a model
guidance_scale
float
1.0-10.0
depends on a model
aspect_ratio
string
"1:1".."21:9"
same as input
Last updated
"options": {
"model": "v1",
"prompt": "add a duck",
"inference_steps": 50
}"options": {
"model": "v1",
"prompt": "add a duck",
"guidance_scale": 4.0
}"options": {
"model": "v1",
"prompt": "add a duck",
"aspect_ratio": "1:1"
}