Models and parameters

01Catalog

The price is per image and depends on the resolution. The amount is charged when the request is accepted, and a failed request is free: the amount goes straight back to your balance.

The model names are identifiers for the model field. GET /models returns the models available to you and your prices: an account can have a personal price or a shorter list.

02Resolution and aspect ratio

• resolution is 1K, 2K, or 4K, case-insensitive. The allowed values depend on the model: only 1K for Nano Banana Flash, Nano Banana 2 Lite, GPT Image 2, and GPT Image 2.5.
• If resolution is omitted, the image is rendered at the default resolution: 2K for Nano Banana Pro, Nano Banana 2, and Nano Banana 2.1, and 1K for Nano Banana Flash, Nano Banana 2 Lite, and all GPT Image models.
• aspect_ratio is an exact value of the form 16:9, with no spaces and no swapped digits. The set of ratios depends on the model.
• If aspect_ratio is omitted, it is 1:1. A missing field does not mean auto.
• Resolution is the model’s size level, not a number of pixels. The width and height fields are not accepted; if your layout needs an exact size, check the dimensions of the file you receive.

What auto means

Nano Banana Pro, Nano Banana 2, Nano Banana 2.1, and Nano Banana Flash can choose the aspect ratio themselves: pass "aspect_ratio": "auto". With reference images, the ratio usually follows the first reference image. Without reference images the shape of the frame is not guaranteed: if your layout depends on it, set aspect_ratio explicitly. Nano Banana 2 Lite and the GPT Image models have no auto value: you set the aspect ratio.

What happens if you choose something impossible

A model that does not support the resolution or aspect ratio you ask for answers 400 before the charge. error.details.supported carries the list of allowed values, and no money is spent.

HTTP/1.1 400 Bad Request

{
  "error": {
    "code": "ERR_INVALID_REQUEST",
    "message": "Unsupported resolution: 4K. Supported: 1K (case-insensitive) (omit resolution to render 1K).",
    "details": { "resolution": "4K", "supported": ["1K"] }
  }
}

HTTP/1.1 400 Bad Request

{
  "error": {
    "code": "ERR_INVALID_REQUEST",
    "message": "Unsupported aspect_ratio: 5:4. Supported: 1:1, 16:9, 9:16, 4:3, 3:4, 2:3, 3:2, 21:9 (omit aspect_ratio to render 1:1).",
    "details": {
      "aspect_ratio": "5:4",
      "supported": [
        "1:1", "16:9", "9:16", "4:3",
        "3:4", "2:3", "3:2", "21:9"
      ]
    }
  }
}

Try it yourself: in the request builder, the grayed-out options show the same refusals.

03Catalog through the API

The method charges nothing and does not take a generation slot. It counts toward the account’s overall rate limit (Limits). The category parameter is optional: pass image for images.

curl --fail-with-body --max-time 20 \
  "https://api.rendergrid.io/api/public/v1/models?category=image" \
  -H "Authorization: Bearer $RENDERGRID_API_KEY"

import os
from decimal import Decimal
import requests

key = os.environ["RENDERGRID_API_KEY"]
response = requests.get(
    "https://api.rendergrid.io/api/public/v1/models",
    params={"category": "image"},
    headers={"Authorization": f"Bearer {key}"},
    timeout=(5, 20),
)
response.raise_for_status()

for model in response.json()["models"]:
    # take the price of a resolution from variants, not from the root price
    prices = {
        v["variant_key"]: Decimal(v["price"])
        for v in model.get("variants", [])
    }
    print(model["id"], prices)

// Node.js 22+, .mjs file
const base = "https://api.rendergrid.io/api/public/v1";
const url = new URL(`${base}/models`);
url.searchParams.set("category", "image");
const key = process.env.RENDERGRID_API_KEY;
const response = await fetch(url, {
  headers: { Authorization: `Bearer ${key}` },
  signal: AbortSignal.timeout(20_000),
});
if (!response.ok) {
  const body = await response.text();
  throw new Error(`HTTP ${response.status}: ${body}`);
}

for (const model of (await response.json()).models) {
  // take the resolution price from variants, not from the root price
  const prices = Object.fromEntries((model.variants ?? []).map((v) => [v.variant_key, v.price]));
  console.log(model.id, prices);
}

{
  "currency": "USD",
  "models": [
    {
      "id": "nano-banana-2",
      "category": "image",
      "display_name": "Nano Banana 2",
      "price": "0.0250",
      "unit": "per_image",
      "variants": [
        { "variant_key": "1k", "price": "0.0250" },
        { "variant_key": "2k", "price": "0.0400" },
        { "variant_key": "4k", "price": "0.0600" }
      ]
    },
    {
      "id": "gpt-image-2.5",
      "category": "image",
      "display_name": "GPT Image 2.5",
      "price": "0.0250",
      "unit": "per_image",
      "variants": [{ "variant_key": "1k", "price": "0.0250" }]
    }
  ]
}

currency

The currency of all amounts in the response: USD.

id

The model identifier for the model field in a generation request.

price

The price of the model’s cheapest resolution, as a string in dollars. It is not the price of your image: that depends on resolution.

unit

The price unit: per_image means per one image.

variants

Prices by resolution: variant_key (1k, 2k, 4k) and price; in the full response a variant also has display_name and unit. Take the price of the chosen resolution from here.

04Price and charge

The price of an image depends on the model and the resolution. For example, Nano Banana 2.1 costs $0.02 at 1K, $0.03 at 2K, and $0.04 at 4K, and GPT Image 2 costs $0.02 at 1K. The lowest price is $0.015 per image at 1K, for Nano Banana Flash and Nano Banana 2 Lite.

If resolution is omitted, the image is rendered and paid for at the default resolution—for Nano Banana Pro, Nano Banana 2, and Nano Banana 2.1 that is 2K, although the root price in the catalog shows 1K. The amount goes into the cost field of the 202 response.

FAQFrequently asked questions