GPT Image
gpt-image-2 — photoreal, prompt-faithful text-to-image and image edit
gpt-image-2 is the default image model: strong at photorealism, faithful to the prompt, and good at
editing an existing image when you pass image_urls. Async task; poll for status.
Endpoint:
POST /api/v2/image/generate｜ Auth:Authorization: Bearer sk_...
Request
| Field | Type | Required | Notes |
|---|---|---|---|
prompt | string | ✅ | Text prompt |
model | string | ⬜ | gpt-image-2 (the default) |
size | string | ⬜ | 1:1 (default) / 16:9 / 9:16 |
image_urls | string[] | ⬜ | Reference images for image-to-image / edit, up to 4 |
seed | int | ⬜ | Random seed |
curl -X POST https://api.aiclonevoicefree.com/api/v2/image/generate \
-H "Authorization: Bearer sk_your_api_key" \
-H "Content-Type: application/json" \
-d '{
"prompt": "a corgi under cherry blossoms, cinematic lighting",
"model": "gpt-image-2",
"size": "16:9"
}'Edit / image-to-image — pass up to 4 reference images:
curl -X POST https://api.aiclonevoicefree.com/api/v2/image/generate \
-H "Authorization: Bearer sk_your_api_key" \
-H "Content-Type: application/json" \
-d '{
"prompt": "put the dog in a santa hat",
"model": "gpt-image-2",
"image_urls": ["https://example.com/dog.png"]
}'Response 202
{ "task_id": "xxxxxxxx", "status": "pending", "capability": "image", "action": "generate", "model": "gpt-image-2" }Getting the result
Poll GET /api/v2/tasks/{task_id}; when done:
{
"status": "completed",
"capability": "image",
"_type": "image.generate",
"imageUrl": "https://.../result.png"
}Billing
Flat 3 video credits per image — independent of size or reference images; n is fixed to 1.
Example: one image (any size, with or without image_urls) = 3 credits.
Settled on completion, failures auto-refunded (see Conventions). Uses video credits.