> ## Documentation Index
> Fetch the complete documentation index at: https://docs.gengen.farm/llms.txt
> Use this file to discover all available pages before exploring further.
# Generate images with Gemini
> Generate or edit images synchronously with Google native image models through the GENGEN Image API.
Use the normalized GENGEN request schema with one of these public model IDs:
* `gemini-3-pro-image`
* `gemini-3.1-flash-image`
* `gemini-3.1-flash-lite-image`
| Property | Value |
| - | - |
| Method | `POST` |
| Endpoint | `/api/gengen/v1/images/generations` |
| Processing | Synchronous |
| Output | Public HTTPS URLs in `outputs.images` |
| URL retention | 24 hours |
## Text-to-image request
```bash theme={"dark"}
curl --request POST \
--url https://gengen.farm/api/gengen/v1/images/generations \
--header 'Authorization: Bearer gengen_live_xxxxxxxxxxxxxxxx' \
--header 'Content-Type: application/json' \
--data '{
"model": "gemini-3.1-flash-image",
"prompt": "A cyan greenhouse at sunrise, editorial photography",
"controls": {
"size": "1K",
"aspectRatio": "16:9"
}
}'
```
## Image editing request
Add public or signed HTTPS URLs, or Base64 data URLs, under `assets.referenceImages`. Raw `gs://` URIs are not accepted; use a signed HTTPS URL instead.
```bash theme={"dark"}
curl --request POST \
--url https://gengen.farm/api/gengen/v1/images/generations \
--header 'Authorization: Bearer gengen_live_xxxxxxxxxxxxxxxx' \
--header 'Content-Type: application/json' \
--data '{
"model": "gemini-3-pro-image",
"prompt": "Keep the product unchanged and replace the background with a clean studio scene.",
"assets": {
"referenceImages": [
"https://example.com/product.png"
]
},
"controls": {
"size": "2K",
"aspectRatio": "1:1"
}
}'
```
## Response
Read generated media from `outputs.images`. Google image responses are URL-only: GENGEN stores each generated image in a dedicated Vercel Blob path and never returns the generated image as Base64. Each URL is retained for 24 hours. The response uses only the normalized top-level fields shown below; duplicate compatibility aliases and provider billing usage are omitted.
```json theme={"dark"}
{
"id": "image:550e8400-e29b-41d4-a716-446655440000",
"model": "gemini-3.1-flash-image",
"status": "succeeded",
"outputs": {
"images": [
"https://example.public.blob.vercel-storage.com/gengen/google-generated-images/2026/08/21/image_550e8400/01-example.png"
],
"expiresAt": "2026-08-22T08:00:00.000Z"
}
}
```
Generated image URLs expire after 24 hours. Download or copy every image to storage you control before `outputs.expiresAt`.
## Parameters
| Parameter | Type | Description |
| - | - | - |
| `model` | string | Required Google native image model ID. |
| `prompt` | string | Generation or editing instruction. Required unless at least one reference image is supplied. |
| `assets.referenceImages` | string\[] | Up to 14 public or signed HTTPS URLs, or Base64 reference images. Raw `gs://` URIs are not accepted. |
| `controls.size` | string | Pro: `1K`, `2K`, `4K`; Nano Banana 2: `512`, `1K`, `2K`, `4K`; Lite: `1K`. |
| `controls.aspectRatio` | string | Supports `1:1`, `2:3`, `3:2`, `3:4`, `4:3`, `4:5`, `5:4`, `9:16`, `16:9`, `21:9`, `1:4`, `4:1`, and `1:8`/`8:1`. Pro and Nano Banana 2 also support `9:21`; Lite does not. |
| `providerOptions.google.generationConfig` | object | Advanced Google image sampling settings such as `temperature` and `topP`. |
| `providerOptions.google.safetySettings` | array | Vertex AI safety settings forwarded to Google. |
GENGEN fixes `candidateCount` to `1`; request multiple alternatives with separate calls. Reference image formats are PNG, JPEG, WebP, HEIC, and HEIF.
Supported sizes vary by model: Nano Banana Pro (Gemini 3 Pro Image) supports `1K`, `2K`, and `4K`; Nano Banana 2 (Gemini 3.1 Flash Image) supports `512`, `1K`, `2K`, and `4K`; Nano Banana 2 Lite (Gemini 3.1 Flash-Lite Image) supports `1K`.
See the [Google model pages](/models/gemini-3.1-flash-image) for detailed capabilities.
## OpenAPI
````yaml openapi/google.yaml POST /images/generations
openapi: 3.1.0
info:
title: Google models API
version: 1.0.0
description: GENGEN endpoints backed by Google Gemini models on Vertex AI.
servers:
- url: https://gengen.farm/api/gengen/v1
description: Production
security:
- bearerAuth: []
paths:
/images/generations:
post:
summary: Generate or edit an image with Gemini
description: >-
Generate or edit an image synchronously with a Google native image
model.
operationId: createGoogleImageGeneration
requestBody:
required: true
content:
application/json:
schema:
$ref: '#/components/schemas/GoogleImageRequest'
example:
model: gemini-3.1-flash-image
prompt: A cyan greenhouse at sunrise, editorial photography
controls:
size: 1K
aspectRatio: '16:9'
responses:
'200':
description: The generated image URLs and their expiry time.
content:
application/json:
schema:
$ref: '#/components/schemas/ImageResult'
default:
$ref: '#/components/responses/Error'
components:
schemas:
GoogleImageRequest:
type: object
required:
- model
anyOf:
- required:
- prompt
- required:
- assets
properties:
model:
type: string
enum:
- gemini-3-pro-image
- gemini-3.1-flash-image
- gemini-3.1-flash-lite-image
description: Google native image model ID.
prompt:
type: string
description: Generation or editing instruction.
assets:
type: object
properties:
referenceImages:
type: array
maxItems: 14
description: >-
Public or signed HTTPS URLs, or Base64 reference images. Raw
`gs://` URIs are not accepted. Google supports at most 14.
items:
type: string
controls:
type: object
properties:
size:
type: string
enum:
- '512'
- 1K
- 2K
- 4K
description: >-
Nano Banana Pro supports 1K/2K/4K; Nano Banana 2 supports
512/1K/2K/4K; Lite supports 1K.
aspectRatio:
type: string
enum:
- '1:1'
- '2:3'
- '3:2'
- '3:4'
- '4:3'
- '4:5'
- '5:4'
- '9:16'
- '16:9'
- '21:9'
- '1:4'
- '4:1'
- '1:8'
- '8:1'
- '9:21'
description: >-
Requested output aspect ratio. `9:21` is supported by Pro and
Nano Banana 2, but not Lite.
providerOptions:
type: object
properties:
google:
type: object
properties:
generationConfig:
type: object
description: >-
Additional Vertex AI generation settings. Output modalities
and candidate count remain fixed by GENGEN.
properties:
temperature:
type: number
topP:
type: number
minimum: 0
maximum: 1
safetySettings:
type: array
items:
type: object
additionalProperties: true
ImageResult:
type: object
required:
- id
- model
- status
- outputs
properties:
id:
type: string
model:
type: string
status:
type: string
const: succeeded
outputs:
type: object
required:
- images
- expiresAt
properties:
images:
type: array
items:
type: string
text:
type: string
expiresAt:
type: string
format: date-time
Error:
type: object
properties:
error:
type: object
properties:
code:
type: string
message:
type: string
responses:
Error:
description: The request could not be completed.
content:
application/json:
schema:
$ref: '#/components/schemas/Error'
securitySchemes:
bearerAuth:
type: http
scheme: bearer
bearerFormat: GENGEN API key
description: A workspace API key beginning with `gengen_live_`.
````