Nano Banana 2 batch vs standard vs Sume: cost of 100 1K images

Google's Gemini API pricing page (read 2026-10-03) lists Gemini 3.1 Flash Image, branded Nano Banana 2, at $0.067 per 1K image on the standard tier. Batch processing is a 50 percent discount: $0.0335. For a 100-image job that is $6.70 standard or $3.35 batch.

On Sume, Nano Banana 2 at the 1K default is $0.08 provider list times 1.25, which is $0.10 per image, so 100 images is $10.00.

All resolutions side by side

The page also lists 0.5K, 2K and 4K. Google's per-image prices follow the number of image tokens each size consumes, at $60 per million tokens on standard.

The 100-image job

At 1K: Google batch is $3.35, Google standard is $6.70, and Sume is $10.00. Sume is about 49 percent above Google standard at 1K, and three times Google batch.

Batch is a trade: Google's page describes it as a discounted tier, and a job in a batch is not meant to come back instantly. If you are producing catalog images overnight, the half-price tier is clearly cheaper. If you need an image inside a user session, only the standard column is realistic.

What Sume adds

The Sume figure buys a single wallet and API key across image, video, speech and assembly, and per the Image API docs failed or cancelled image generations are not billed. Whether that is worth $3.30 per 100 images over Google standard depends on how much of your pipeline already runs on Sume.

Use this snippet to price your own batch at each column.

rates = {'google batch': 0.0335, 'google standard': 0.067, 'sume': 0.10}
images = 100
for name, per_image in rates.items():
    print(f'{name}: ${per_image * images:.2f} for {images} images')