Pick one of four: an n=4 call costs $0.60 or $0.22 on Sume at 2K

The short answer

An image call with n set to 4 at 2K costs $0.60 on Sume with Nano Banana 2.1 (4 x $0.15) and $0.22 with GPT Image 2.5 medium (4 x $0.055625 = $0.2225). The docs state the rule directly: you pay cost_usd times n.

That rule makes n a pure multiplier, so the choice between four calls and one call with n at 4 is about latency and handling, not price.

The arithmetic

Four single calls cost the same as one call with n at 4. The difference is that n repeats the same prompt and inputs, while separate calls let you vary the prompt between takes.

Use n when you want variety from one prompt, and separate calls when each take tests a different idea.

How to run it

A larger n takes longer, and the docs warn that slow configurations, including a large n, are the most likely to degrade from 200 to a 202 job envelope. Write the 202 branch before you raise n.

Read the n range for each model in GET /v1/images/models. The generic limit is 10, but a model's own ceiling can be lower.

Retries and what to watch

The multiplier rule gives a quick way to size a request before you send it: unit price times n. For a 4K call the same rule gives $2.00 for ten Nano Banana 2.1 images and $2.225 for ten GPT Image 2.5 high images.

These are the checks that keep the multiplier from surprising you.

• Read the n range for the model, because the documented ceiling is 10 and a model may allow fewer.
• Sum usage.cost across calls to confirm the bill matches unit price times n.

What the docs say to check

The endpoint pricing lines are what Sume charges your wallet and already include the Sume margin, so a call costs cost_usd times n. Sume bills image generation all or nothing: a generation that fails or is cancelled is not charged.

Check the status code, not the body shape. POST /v1/images waits up to 30 seconds and returns the images with 200. Slow settings (4K, high quality, a large n) are the most likely to come back as 202 with a job envelope, and you then read the images from the job result endpoint described in Jobs and results.

Rounding: totals are shown to the cent, so the GPT n=1 call reads as $0.06 while the price is $0.055625 per image.

Every price in this post is a snapshot dated 2026-10-08. The Sume figures come from the public catalog and the Image API docs, the Google figures come from its Gemini API pricing page, and no figure here is a benchmark of quality or speed. Prices and model lists change, so confirm them on the day you spend money.