Image Generation and Editing

Image APIs support generation and editing. Exact sizes, quality values, formats and streaming support depend on the model and upstream channel.

Endpoints

Method	Path	Description
`POST`	`/v1/images/generations`	Generate images
`POST`	`/v1/images/edits`	Edit images, with JSON or multipart form input
`POST`	`/v1/edits`	Compatibility path. Prefer `/v1/images/edits`

Request Fields

Field	Type	Required	Description
`model`	string	Yes	Image model ID
`prompt`	string	Required for generation	Image prompt
`n`	integer	No	Number of images, default 1
`size`	string	No	Image size such as `1024x1024`. Do not use the multiplication sign `×`
`quality`	string	No	Quality such as `standard`, `hd` or `auto`
`response_format`	string	No	Response format such as `url` or `b64_json`, depending on model
`style`	any	No	OpenAI image extension field
`background`, `moderation`	any	No	Newer image model extension fields
`output_format`, `output_compression`	any	No	Output format and compression options
`partial_images`, `stream`	any	No	Used by image models that support streaming
`image`, `images`	file/string/array	Required for edits	Images to edit. Multipart accepts `image`, `image[]` or `image[n]`
`mask`	file/string	No	Edit mask
`input_fidelity`	any	No	Input fidelity, depending on model
`watermark`	boolean	No	Watermark switch, depending on channel support

Generate an Image

curl https://api.tensoraxis.ai/v1/images/generations \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $TENSORAXIS_API_KEY" \
  -d '{
    "model": "gpt-image-1",
    "prompt": "A minimalist poster of a futuristic city",
    "size": "1024x1024"
  }'

Edit an Image

Use multipart form data when uploading local files:

curl https://api.tensoraxis.ai/v1/images/edits \
  -H "Authorization: Bearer $TENSORAXIS_API_KEY" \
  -F model="gpt-image-1" \
  -F prompt="Change the background to a night city" \
  -F image="@input.png"

When the selected model supports it, JSON image URLs can also be used:

{
  "model": "gpt-image-1",
  "prompt": "Change the background to a night city",
  "image": "https://example.com/input.png"
}

Response Shape

{
  "created": 1710000000,
  "data": [
    {
      "url": "https://example.com/result.png",
      "b64_json": "",
      "revised_prompt": "..."
    }
  ]
}

Video generation is not an image API. For Seedance / Doubao video models, use the async task body in Video Generation.