One prompt, one image.
You send a prompt to POST /v1/images/generations and get the image in the same response, as a link or as Base64. The call follows the OpenAI format.
These models are available.
The table reads the same catalog as the pricing page. Prices in euros per image, from the cheapest size and quality.
| Model | Sizes | Quality | Editing | From€ per image |
|---|---|---|---|---|
google/gemini-3.1-flash-image
Google
|
square, portrait, landscape, 2k |
medium |
no | 0.0809 |
google/gemini-nano-banana-2.1
Google
|
square, portrait, landscape, 2k |
medium |
no | 0.0626 |
ionos/black-forest-labs/FLUX.2-klein-4B
IONOS
|
square, portrait, landscape, 2k |
medium |
yes | 0.0159 Editing from 0.0171 |
minimax/h3-image
MiniMax
|
square, portrait, landscape, 2k |
medium |
yes | 0.0296 Editing from 0.0365 |
minimax/h3-image-highspeed
MiniMax
|
square, portrait, landscape, 2k |
medium |
yes | 0.0444 Editing from 0.0592 |
openai/gpt-image-2.6-sunburst
OpenAI
|
square, portrait, landscape, 2k |
low, medium |
yes | 0.0376 Editing from 0.0513 |
Sizes by name instead of pixels.
In the size field you name one of the four sizes. Which ones a model supports is shown in the table above.
square- square
portrait- portrait
landscape- landscape
2k- landscape, in higher resolution
A fixed price per image.
Before the call, FlowModels reserves the price of the image. It is charged as soon as the image is stored. If the generation fails, the reservation is released and you pay nothing.
- One image per request
- Every request generates exactly one image,
nmust be 1. - Editing
POST /v1/images/editschanges an existing image if the model supports editing. The image comes asimage_urlorimage_file_id, as JPEG, PNG or WebP up to 10 MiB.- Retention
- Images stay available for 7 days.
GET /v1/images/{id}/contentredirects to a download link that is valid for 15 minutes, andDELETE /v1/images/{id}deletes sooner.
Request and response.
With response_format you choose between url and b64_json. The response carries the ID of the image, under which you can fetch it again later.
curl https://api.flowmodels.one/v1/images/generations \
-H "Authorization: Bearer $FLOWMODELS_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "google/gemini-3.1-flash-image",
"prompt": "An oak dining table in a bright workshop",
"size": "landscape",
"response_format": "url"
}'
{
"id": "018fa123-4567-7abc-9def-0123456789ab",
"object": "image",
"created": 1794560400,
"model": "google/gemini-3.1-flash-image",
"status": "completed",
"data": [
{ "url": "https://api.flowmodels.one/v1/images/018fa123-4567-7abc-9def-0123456789ab/content" }
],
"content_type": "image/png",
"expires_at": "2026-11-20T10:00:00+01:00",
"flowai_request_id": "01HXYZ8M3F6V2KQ"
}
| Error code | Meaning |
|---|---|
422 image_content_rejected |
The image was rejected because of its content. |
422 image_request_rejected |
The model rejected the request. |
503 image_provider_unavailable |
The model cannot be reached right now. The reservation is released. |
503 image_pricing_unavailable |
There is no price for this setting right now. Nothing is charged. |
Try it in the dashboard first.
Under Images you try out prompt, size and quality, see the price per image before you start and find your latest images in a gallery. Billing runs through the API key you pick.