Edit Images with VoidAI - VoidAI

POST https://api.voidai.app/v1/images/edits Send multipart/form-data with authentication. Image routes are unavailable to free-plan accounts. Select a model that advertises /v1/images/edits; the default is gpt-image-2.

Request fields

Field	Type	Description
`image` or `image[]`	file(s), required	One file named `image`, or repeated `image[]` fields; 1–16 input files.
`prompt`	string, required	Non-whitespace edit instructions; current GPT image models allow at most 32,000 characters.
`model`	string	Defaults to `gpt-image-2`.
`mask`	file	Optional mask where supported by the model/provider.
`n`	integer	1–10 output images.
`size`, `quality`, `output_format`, `output_compression`	string / integer	Same gateway values and model constraints as image generation.
`background`, `moderation`, `seed`, `stream`, `partial_images`, `user`	varies	Same constraints as image generation.
`input_fidelity`	string	`high` or `low`; rejected by `gpt-image-2`.

The editing form does not accept generation-only thinking, style, or response_format options. gpt-image-2 does not support transparent backgrounds. File format and upload-size constraints depend on the selected provider; the gateway does not define a universal PNG-only or 4 MB limit.

curl https://api.voidai.app/v1/images/edits \
  -H "Authorization: Bearer $VOIDAI_API_KEY" \
  -F model=gpt-image-2 \
  -F image=@source.png \
  -F 'prompt=Make the sky a warm sunset' \
  -F n=1 \
  -F size=1024x1024 \
  -F output_format=png

For multiple references, repeat -F 'image[]=@reference.png' using each filename. Let your HTTP client set the multipart boundary rather than setting a plain Content-Type header manually. Responses and per-image billing follow image generation. Use non-streaming unless your selected model/provider’s streaming support is confirmed.