Klingobanana API

Base URL: https://klingobanana.com. All requests and responses are JSON.

An active subscription is required. There is no free tier and no free credits: every request spends credits, and credits come from a subscription plan or a top-up. API access is also enabled per account by our team; contact us to request it. Once enabled, create keys in your account under the API tab (up to 5 active keys).

The API covers images only. Video generation is not available through the API.

Send your key as a bearer token in the Authorization header. Keys begin with kb_live_.

Authorization: Bearer kb_live_xxxxxxxxxxxxxxxxxxxxxxxx

Keep keys server-side. A revoked key stops working immediately, as does removing API access from the account.

1. Submit a generation request. You receive a job id and the model id. Credits are charged at this point.
2. Poll the status endpoint with that id until the status is a terminal value. A reasonable interval is 3 seconds.
3. On success the response contains result URLs. Download them promptly and store your own copies.

Image jobs are produced while the submit request is open, so the submit call can take several seconds and the first poll will usually already return the result.

Single-image generation or edit with a fixed model pair. Text-to-image by default; pass imageUrl to edit an existing image.

• prompt (string, required): 1 to 2000 characters.
• imageUrl (string, optional): an https:// URL or a data:image/ URI. Switches to edit mode.
• aspectRatio (string, optional, default 1:1): one of 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 4:5, 5:4.

curl -X POST https://klingobanana.com/api/v1/images/generations \
  -H "Authorization: Bearer $KB_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt": "A ceramic teapot on a linen cloth, soft window light", "aspectRatio": "4:5"}'

{
  "id": "rw:0b6f4e0e-5b1c-4c52-9a43-1f0a2f9b7c11",
  "model": "fal-ai/gemini-3-pro-image-preview",
  "status": "processing"
}

Poll a job created above. The model query parameter is required: pass back the model value from the submit response.

curl "https://klingobanana.com/api/v1/images/generations/$ID?model=fal-ai/gemini-3-pro-image-preview" \
  -H "Authorization: Bearer $KB_API_KEY"

{ "id": "rw:...", "status": "processing", "progress": 5 }

{
  "id": "rw:...",
  "status": "completed",
  "images": [{ "url": "https://firebasestorage.googleapis.com/...", "width": 1024, "height": 1280 }]
}

{ "id": "rw:...", "status": "failed", "error": "..." }

Lower-level image endpoint where you choose the model. Poll with GET /api/v1/status.

• prompt (string, required): 1 to 2000 characters.
• modelId (string, required): a model from the pricing tables below.
• referenceImageUrl, faceImageUrl (string or null, optional): reference images, used only by models that accept image input.
• aspectRatio (string, default 9:16).
• numImages (integer 1 to 8, default 1).
• negativePrompt (string), numInferenceSteps (integer 1 to 100), seed (integer), sizeMultiplier (number 0.5 to 2, default 1): optional tuning fields.

curl -X POST https://klingobanana.com/api/v1/generate/image \
  -H "Authorization: Bearer $KB_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt": "Studio portrait, 85mm, shallow depth of field", "modelId": "fal-ai/nano-banana-2", "aspectRatio": "4:5"}'

{
  "requestId": "rw:0b6f4e0e-5b1c-4c52-9a43-1f0a2f9b7c11",
  "modelId": "fal-ai/nano-banana-2"
}

Query parameters modelId and requestId are both required.

curl "https://klingobanana.com/api/v1/status?modelId=fal-ai/nano-banana-2&requestId=$REQUEST_ID" \
  -H "Authorization: Bearer $KB_API_KEY"

{ "status": "processing", "progress": 5, "queuePosition": null }

{
  "status": "done",
  "results": [{ "url": "https://firebasestorage.googleapis.com/...", "mediaType": "image", "width": 1024, "height": 1280 }]
}

{ "status": "failed", "error": "..." }

Limits apply per API key. Exceeding any of them returns 429 with a Retry-After header (seconds).

• 30 requests per minute, including status polling.
• 3 concurrent generations in flight at once.
• Daily credit cap (default 500 credits per UTC day).

Errors return a JSON body of the form { "error": "message" } with the HTTP status below.

Each request is charged in credits when it is submitted, and refunded automatically if the job fails.

POST /api/v1/images/generations uses a fixed model pair, priced per request:

POST /api/v1/generate/image accepts any of the models below via modelId. Cost is charged per image, multiplied by numImages (Gemini and GPT image models always produce one image).

Credits come from your subscription: starter includes 100 per cycle, pro includes 300 per cycle, business includes 900 per cycle. Top-ups are available in the studio.