Nano Banana multiple images at once: n range vs Gemini's count

Google says the Gemini model won't always follow the exact number of images a user asks for in the prompt. On Sume, set the n parameter instead: it is a counted request field, up to 10 per call in the docs, with a lower ceiling on most models. Read the n range from the catalog before you pick a number.

Google's limitation is from its image generation page; Sume's from Image generation, read 2026-10-01.

Why does asking in the prompt not guarantee a count?

Under "Limitations", Google writes that the model won't always follow the exact number of image outputs that the user explicitly asks for. A prompt like "make four variations" is guidance to the model, not a parameter, so treat the count as approximate and check what came back.

How does n work on Sume?

The docs say to request up to 10 images per call with n, and that per-model ceilings are lower. The API contract notes that every other image model takes n as a 1 to 4 range, and the catalog includes google/nano-banana-pro. A value outside a model's range is not listed for it, and unlisted settings return 400 unsupported_parameter.

What does a request look like?

Authenticate with your API key, pick a model and set n.

const res = await fetch("https://api.sume.com/v1/images", {
  method: "POST",
  headers: {
    Authorization: "Bearer " + process.env.SUME_API_KEY,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    model: "google/nano-banana-pro",
    prompt: "a ceramic mug on a linen cloth, soft window light",
    n: 4,
  }),
});
console.log(res.status, (await res.json()).data?.length);

What if the request takes too long?

The docs note that slow configurations, including large n, are likelier to degrade to a 202 job envelope. Check the status code: 200 is the image response, 202 is a job to poll. See multiple images with `n` for the same behavior on ChatGPT Image 2.5.