12-banner ad pack in 4:1, 1:4 and 8:1: $1.20 at 1K on Sume

A pack of 12 banners (three ratios, four options each) on Nano Banana 2.1 costs $1.20 at 1K, $0.90 at 0.5K, $1.80 at 2K and $2.40 at 4K on Sume. Three calls with n: 4 produce it, one per ratio, because the price is per image and independent of the ratio.

Cost of the pack

Per-image prices are from the price tool; totals are 12 times the price (read 2026-10-09).

Send the three calls

Docs state that cost_usd is per image and you pay cost_usd x n. The per-call usage.cost should match the table. Read the n range from the model's descriptor; Sume allows up to 10 in the schema, with lower per-model ceilings.

import os, requests

H = {"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"}
for ratio in ["4:1", "1:4", "8:1"]:
    r = requests.post("https://api.sume.com/v1/images", headers=H, timeout=60, json={
        "model": "google/nano-banana-2.1",
        "prompt": "Autumn sale banner for a coffee shop, warm light, space for a headline",
        "aspect_ratio": ratio,
        "resolution": "1K",
        "n": 4,
    })
    print(ratio, r.status_code, r.json().get("usage", {}).get("cost"))

Draft cheap, finalize the keepers

A pack at 0.5K costs $0.90. Re-running only the three keepers at 2K adds $0.45, for $1.35 total, against $1.80 for all twelve at 2K.

Gotchas

• Do not send these ratios to Nano Banana Pro or sume/auto; neither lists them.
• Large n at 4K is the likeliest combination to degrade to 202, so handle both status codes.
• A failed generation is not billed, so a retried call does not double the cost.