Nano Banana 2.1 or Pro for ad creative? 2.1 at 2K beats Pro at 1K

For ad creative on Sume, Nano Banana 2.1 is the cheaper and wider choice on paper: it bills $0.10 at 1K and $0.15 at 2K, against $0.1875 for Nano Banana Pro at those tiers, and it lists 15 aspect ratios against Pro's 11. Pro is more expensive at 4K as well, at $0.375 against $0.20. The catalog says nothing about picture quality, so the only fair claim is about price, ratios and limits.

The numbers below come from the catalog code and Image API docs read on 2026-10-10. Google's image generation docs, read the same day, describe Nano Banana 2.1's own features such as thinking levels and up to 10 object and 4 character references.

Price by tier

The pricing code scales Nano Banana 2.1 at 0.5K, 1K, 2K and 4K list rates of $0.06, $0.08, $0.12 and $0.16, then bills list times 1.25. Pro has two list rates in the code: $0.15 for every tier but 4K, and $0.30 at 4K.

What a 24-ad set costs

Say you test 24 ad images, four concepts times six formats. On 2.1 at 2K that is 24 x $0.15 = $3.60. On Pro at 2K it is 24 x $0.1875 = $4.50. At 4K the figures are $4.80 and $9.00. If you draft at 0.5K on 2.1 first, 24 drafts cost $1.80, and re-rendering the six winners at 4K adds 6 x $0.20 = $1.20 for a total of $3.00.

The gap is small for one test and large for a catalog. A thousand 4K finals cost $200 on 2.1 and $375 on Pro.

Ratios and limits that decide it

Ad placements are not all 1:1 and 9:16. Nano Banana 2.1 lists 4:1, 1:4, 8:1 and 1:8 on top of Pro's set (auto, 21:9, 16:9, 3:2, 4:3, 5:4, 1:1, 4:5, 3:4, 2:3, 9:16), so a leaderboard banner or a tall strip is possible on 2.1 and not on Pro. Both rows list up to 4 outputs per call, up to 10 references and png, jpeg or webp output.

Neither row lists quality, mask_url or background, and Sume rejects them with a 400 rather than ignoring them. If your creative needs a masked edit or a transparent PNG, those fields belong to ChatGPT Image 2.5 in the catalog.

• Pick 2.1 when cost and extreme banner ratios matter.
• Pick Pro only if you have compared outputs on your own brief and prefer them, since the premium is the whole difference in the catalog.
• Pin the model id; sume/auto hides which family ran.

A simple decision rule

Run your three hardest briefs on both rows at 2K and look at them side by side. If 2.1 holds up, you save $0.0375 per image. If you can name a defect that only Pro avoids, the premium has a reason. Do not decide from a spec table alone.

Where Pro could still be worth it

The catalog cannot say whether Pro looks better on your product, and this post does not claim it. What it can say is that the premium is fixed and known: $0.0375 more per image at every tier up to 2K, and $0.175 more at 4K. A campaign that ships 200 hero images a month pays $7.50 extra at 2K and $35 extra at 4K, which is small next to a photo shoot and large next to a draft run.

So use the two rows differently. Draft broad on the cheaper row, shortlist, and spend the premium only on the final frames where you can show a difference to the person approving the work. If you cannot show a difference, do not pay for it.

Both rows are billed per output image, so asking for four variants in one call multiplies the figures above by four. Check the wallet charge as the endpoint price times n before a large run.