Google Flow
Image & Video API

OpenAI-compatible API for Google's Imagen 4, Nano Banana 2, Nano Banana Pro image generation and Veo 2/3.1 video generation. Supports text-to-image, image-to-image, text-to-video, image-to-video, and reference-based video generation.

Nano Banana 2 (NARWHAL) Nano Banana Pro (GEM_PIX_2) Imagen 4 (IMAGEN_3_5) Veo 2.0 / 2.1 / 3.1

Endpoints

GET/v1/models GET/v1/models/aliases POST/v1/chat/completions POST/models/{model}:generateContent POST/models/{model}:streamGenerateContent POST/api/tokens GET/api/tokens GET/api/system/info

GET /v1/models List available models ▶

Returns all available image and video generation models with their IDs and descriptions.

Request

curl BASE_URL/v1/models \
  -H "Authorization: Bearer YOUR_API_KEY"

Response

{
  "object": "list",
  "data": [
    {
      "id": "gemini-3.1-flash-image-landscape",
      "object": "model",
      "owned_by": "flow2api",
      "description": "Image generation - NARWHAL"
    },
    ...
  ]
}

GET /v1/models/aliases Simplified model aliases ▶

List simplified model aliases for generationConfig-based resolution. Useful when using Gemini-native endpoints.

Request

curl BASE_URL/v1/models/aliases \
  -H "Authorization: Bearer YOUR_API_KEY"

POST /v1/chat/completions OpenAI-compatible generation ▶

The primary endpoint. OpenAI-compatible unified generation for images and videos. Select a model to control what gets generated. Supports streaming responses.

Parameter	Type	Description
model required	string	Model ID (e.g. `gemini-3.1-flash-image-landscape`)
messages required	array	Array of `{ role, content }` messages. Content can be a string (text prompt) or array of content parts for multimodal input.
stream optional	boolean	Enable streaming response. Default: `false`
image optional	string	Base64-encoded image for image-to-image generation
video optional	string	Base64-encoded video for video reference

Text-to-Image

curl -X POST BASE_URL/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3.1-flash-image-landscape",
    "messages": [
      {
        "role": "user",
        "content": "A red apple on a wooden table, studio lighting, minimal background"
      }
    ],
    "stream": true
  }'

Image-to-Image (edit an existing image)

curl -X POST BASE_URL/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3.1-flash-image-landscape",
    "messages": [
      {
        "role": "user",
        "content": [
          {
            "type": "text",
            "text": "Turn this into a watercolor painting style"
          },
          {
            "type": "image_url",
            "image_url": {
              "url": "data:image/jpeg;base64,<BASE64_IMAGE>"
            }
          }
        ]
      }
    ],
    "stream": true
  }'

Text-to-Video

curl -X POST BASE_URL/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "veo_3_1_t2v_fast_landscape",
    "messages": [
      {
        "role": "user",
        "content": "A kitten chasing butterflies across a sunlit meadow"
      }
    ],
    "stream": true
  }'

Image-to-Video (first/last frame)

curl -X POST BASE_URL/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "veo_3_1_i2v_s_fast_fl",
    "messages": [
      {
        "role": "user",
        "content": [
          {
            "type": "text",
            "text": "Transition from the first frame to the second"
          },
          {
            "type": "image_url",
            "image_url": {
              "url": "data:image/jpeg;base64,<FIRST_FRAME_BASE64>"
            }
          },
          {
            "type": "image_url",
            "image_url": {
              "url": "data:image/jpeg;base64,<LAST_FRAME_BASE64>"
            }
          }
        ]
      }
    ],
    "stream": true
  }'

Multi-Reference Video (R2V)

curl -X POST BASE_URL/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "veo_3_1_r2v_fast",
    "messages": [
      {
        "role": "user",
        "content": [
          {
            "type": "text",
            "text": "Create a video featuring these reference elements"
          },
          {
            "type": "image_url",
            "image_url": { "url": "data:image/jpeg;base64,<REF1>" }
          },
          {
            "type": "image_url",
            "image_url": { "url": "data:image/jpeg;base64,<REF2>" }
          }
        ]
      }
    ],
    "stream": true
  }'

Response (streaming SSE)

data: {"id":"chatcmpl-...","object":"chat.completion.chunk","model":"gemini-3.1-flash-image-landscape","choices":[{"index":0,"delta":{"role":"assistant","content":"![image](data:image/png;base64,iVBORw0KGgo...)"},"finish_reason":"stop"}]}

data: [DONE]

POST /models/{model}:generateContent Gemini-native generation ▶

Gemini-native endpoint. Supports systemInstruction, contents[].parts[].text, inlineData, fileData, and generationConfig with responseModalities and imageConfig. Also available at /v1beta/models/{model}:generateContent.

Parameter	Type	Description
contents required	array	Array of content objects with `role` and `parts`
systemInstruction optional	object	System instruction with `parts[].text`
generationConfig optional	object	`responseModalities` (e.g. `["IMAGE"]`), `imageConfig.aspectRatio` (e.g. `"1:1"`), `imageConfig.imageSize` (e.g. `"1K"`, `"2K"`, `"4K"`)

Gemini-native Image Generation

curl -X POST BASE_URL/models/gemini-3.1-flash-image:generateContent \
  -H "x-goog-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "systemInstruction": {
      "parts": [{ "text": "Return an image only." }]
    },
    "contents": [
      {
        "role": "user",
        "parts": [
          { "text": "A red apple on a wooden table, studio lighting" }
        ]
      }
    ],
    "generationConfig": {
      "responseModalities": ["IMAGE"],
      "imageConfig": {
        "aspectRatio": "1:1",
        "imageSize": "1K"
      }
    }
  }'

Response

{
  "candidates": [
    {
      "content": {
        "parts": [
          {
            "inlineData": {
              "mimeType": "image/png",
              "data": "iVBORw0KGgo..."
            }
          }
        ],
        "role": "model"
      }
    }
  ]
}

POST /models/{model}:streamGenerateContent Gemini streaming generation ▶

Same as generateContent but returns a streaming response. Also available at /v1beta/models/{model}:streamGenerateContent. Append ?alt=sse for SSE format.

Request

curl -X POST "BASE_URL/models/gemini-3.0-pro-image:streamGenerateContent?alt=sse" \
  -H "x-goog-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "contents": [
      {
        "role": "user",
        "parts": [{ "text": "Generate an artistic portrait" }]
      }
    ],
    "generationConfig": {
      "responseModalities": ["IMAGE"],
      "imageConfig": { "aspectRatio": "3:4", "imageSize": "2K" }
    }
  }'

POST /api/tokens Add a Google session token ▶

Add a new Google session token (ST — __Secure-next-auth.session-token cookie value from labs.google). The service will automatically obtain an access token (AT) from it. Requires admin authentication.

Parameter	Type	Description
st required	string	The session token cookie value
remark optional	string	A label/note for this token
image_enabled optional	boolean	Enable image generation for this token. Default: `true`
video_enabled optional	boolean	Enable video generation for this token. Default: `true`

Request (Admin auth required — login first at /manage)

curl -X POST BASE_URL/api/tokens \
  -H "Content-Type: application/json" \
  -b "session=ADMIN_SESSION_COOKIE" \
  -d '{
    "st": "YOUR_GOOGLE_SESSION_TOKEN",
    "remark": "Main account"
  }'

GET /api/tokens List all tokens ▶

List all registered Google session tokens and their status. Requires admin authentication.

Request

curl BASE_URL/api/tokens \
  -b "session=ADMIN_SESSION_COOKIE"

GET /api/system/info System information ▶

Returns system information including version, uptime, token count, and configuration. Requires admin authentication.

Request

curl BASE_URL/api/system/info \
  -b "session=ADMIN_SESSION_COOKIE"

Image Generation Models

Nano Banana 2 (NARWHAL) — gemini-3.1-flash-image

Model ID	Orientation	Resolution
gemini-3.1-flash-image-landscape	Landscape	Standard
gemini-3.1-flash-image-portrait	Portrait	Standard
gemini-3.1-flash-image-square	Square	Standard
gemini-3.1-flash-image-four-three	4:3	Standard
gemini-3.1-flash-image-three-four	3:4	Standard
gemini-3.1-flash-image-landscape-2k	Landscape	2K
gemini-3.1-flash-image-portrait-2k	Portrait	2K
gemini-3.1-flash-image-square-2k	Square	2K
gemini-3.1-flash-image-four-three-2k	4:3	2K
gemini-3.1-flash-image-three-four-2k	3:4	2K
gemini-3.1-flash-image-landscape-4k	Landscape	4K
gemini-3.1-flash-image-portrait-4k	Portrait	4K
gemini-3.1-flash-image-square-4k	Square	4K
gemini-3.1-flash-image-four-three-4k	4:3	4K
gemini-3.1-flash-image-three-four-4k	3:4	4K

Nano Banana Pro (GEM_PIX_2) — gemini-3.0-pro-image

Model ID	Orientation	Resolution
gemini-3.0-pro-image-landscape	Landscape	Standard
gemini-3.0-pro-image-portrait	Portrait	Standard
gemini-3.0-pro-image-square	Square	Standard
gemini-3.0-pro-image-four-three	4:3	Standard
gemini-3.0-pro-image-three-four	3:4	Standard
gemini-3.0-pro-image-landscape-2k	Landscape	2K
gemini-3.0-pro-image-portrait-2k	Portrait	2K
gemini-3.0-pro-image-square-2k	Square	2K
gemini-3.0-pro-image-four-three-2k	4:3	2K
gemini-3.0-pro-image-three-four-2k	3:4	2K
gemini-3.0-pro-image-landscape-4k	Landscape	4K
gemini-3.0-pro-image-portrait-4k	Portrait	4K
gemini-3.0-pro-image-square-4k	Square	4K
gemini-3.0-pro-image-four-three-4k	4:3	4K
gemini-3.0-pro-image-three-four-4k	3:4	4K

Imagen 4 (IMAGEN_3_5) — imagen-4.0-generate-preview

Model ID	Orientation	Resolution
imagen-4.0-generate-preview-landscape	Landscape	Standard
imagen-4.0-generate-preview-portrait	Portrait	Standard

Gemini Flash (GEM_PIX) — gemini-2.5-flash-image

Model ID	Orientation	Resolution
gemini-2.5-flash-image-landscape	Landscape	Standard
gemini-2.5-flash-image-portrait	Portrait	Standard

Video Generation Models

Text-to-Video (T2V)

Model ID	Orientation	Quality
veo_3_1_t2v_fast_landscape	Landscape	Fast
veo_3_1_t2v_fast_portrait	Portrait	Fast
veo_3_1_t2v_landscape	Landscape	Standard
veo_3_1_t2v_portrait	Portrait	Standard
veo_3_1_t2v_fast_ultra	Landscape	Ultra
veo_3_1_t2v_fast_portrait_ultra	Portrait	Ultra
veo_3_1_t2v_fast_ultra_relaxed	Landscape	Ultra Relaxed
veo_3_1_t2v_fast_portrait_ultra_relaxed	Portrait	Ultra Relaxed
veo_2_1_fast_d_15_t2v_landscape	Landscape	Veo 2.1
veo_2_1_fast_d_15_t2v_portrait	Portrait	Veo 2.1
veo_2_0_t2v_landscape	Landscape	Veo 2.0
veo_2_0_t2v_portrait	Portrait	Veo 2.0

Image-to-Video (I2V) — First/Last Frame

1 image = first frame only. 2 images = first frame + last frame transition.

Model ID	Orientation	Quality
veo_3_1_i2v_s_fast_fl	Landscape	Fast
veo_3_1_i2v_s_fast_portrait_fl	Portrait	Fast
veo_3_1_i2v_s_fast_ultra_fl	Landscape	Ultra
veo_3_1_i2v_s_fast_portrait_ultra_fl	Portrait	Ultra
veo_3_1_i2v_s_fast_ultra_relaxed	Landscape	Ultra Relaxed
veo_3_1_i2v_s_fast_portrait_ultra_relaxed	Portrait	Ultra Relaxed
veo_3_1_i2v_s_landscape	Landscape	Standard
veo_3_1_i2v_s_portrait	Portrait	Standard
veo_2_1_fast_d_15_i2v_landscape	Landscape	Veo 2.1
veo_2_1_fast_d_15_i2v_portrait	Portrait	Veo 2.1
veo_2_0_i2v_landscape	Landscape	Veo 2.0
veo_2_0_i2v_portrait	Portrait	Veo 2.0

Reference-to-Video (R2V) — Multiple Reference Images

Up to 3 reference images. System auto-selects model key based on image count.

Model ID	Orientation	Quality
veo_3_1_r2v_fast	Landscape	Fast
veo_3_1_r2v_fast_portrait	Portrait	Fast
veo_3_1_r2v_fast_ultra	Landscape	Ultra
veo_3_1_r2v_fast_portrait_ultra	Portrait	Ultra
veo_3_1_r2v_fast_ultra_relaxed	Landscape	Ultra Relaxed
veo_3_1_r2v_fast_portrait_ultra_relaxed	Portrait	Ultra Relaxed

Video Upscale Models

Upscale generated videos to 1080p or 4K resolution.

Model ID	Source	Output
veo_3_1_t2v_fast_4k	T2V	4K
veo_3_1_t2v_fast_portrait_4k	T2V	4K
veo_3_1_t2v_fast_ultra_4k	T2V Ultra	4K
veo_3_1_t2v_fast_portrait_ultra_4k	T2V Ultra	4K
veo_3_1_t2v_fast_1080p	T2V	1080p
veo_3_1_t2v_fast_portrait_1080p	T2V	1080p
veo_3_1_t2v_fast_ultra_1080p	T2V Ultra	1080p
veo_3_1_t2v_fast_portrait_ultra_1080p	T2V Ultra	1080p
veo_3_1_i2v_s_fast_ultra_fl_4k	I2V Ultra	4K
veo_3_1_i2v_s_fast_portrait_ultra_fl_4k	I2V Ultra	4K
veo_3_1_i2v_s_fast_ultra_fl_1080p	I2V Ultra	1080p
veo_3_1_i2v_s_fast_portrait_ultra_fl_1080p	I2V Ultra	1080p
veo_3_1_r2v_fast_ultra_4k	R2V Ultra	4K
veo_3_1_r2v_fast_portrait_ultra_4k	R2V Ultra	4K
veo_3_1_r2v_fast_ultra_1080p	R2V Ultra	1080p
veo_3_1_r2v_fast_portrait_ultra_1080p	R2V Ultra	1080p

Important Notes

All generation endpoints use streaming ("stream": true) for best results. The response is delivered via Server-Sent Events (SSE).
Image models return base64-encoded images in the response content. Video models return base64-encoded MP4 video data.
The Google session token (ST) expires periodically. Update it via the Admin Panel or use the browser extension Flow2API-Token-Updater for automatic refresh.
Image-to-Video models support 1 image (first frame) or 2 images (first + last frame transition).
Reference-to-Video (R2V) models support up to 3 reference images.
reCAPTCHA Enterprise is solved automatically by the service using the configured captcha method (browser/yescaptcha/personal).
Multiple Google tokens are supported for load balancing. Add more tokens in the Admin Panel for higher throughput.
Concurrency is controlled per-token and globally. The service will queue requests that exceed the configured concurrency limit.
Video upscale models (4K/1080p) generate higher resolution versions of an existing video generation.

Google FlowImage & Video API

Nano Banana 2 (NARWHAL) — gemini-3.1-flash-image

Nano Banana Pro (GEM_PIX_2) — gemini-3.0-pro-image

Imagen 4 (IMAGEN_3_5) — imagen-4.0-generate-preview

Gemini Flash (GEM_PIX) — gemini-2.5-flash-image

Text-to-Video (T2V)

Image-to-Video (I2V) — First/Last Frame

Reference-to-Video (R2V) — Multiple Reference Images

Video Upscale Models

Important Notes

Google Flow
Image & Video API