Nano Banana Pro vs 2.1 ratios: four strips only 2.1 lists, at $0.10

Nano Banana 2.1 lists four aspect ratios that Nano Banana Pro does not: 4:1, 1:4, 8:1 and 1:8. It also costs less on the endpoint pricing line, $0.10 against $0.1875 for Pro. Both rows take 1 to 4 images per call, up to 10 references, png, jpeg or webp, and the four resolution tiers 512, 1K, 2K and 4K. So for a strip banner or a tall scroll image, 2.1 is the only choice of the pair, and for any other shape the contract does not separate them.

All figures are from the Sume catalog on origin/main, read 2026-10-08. The endpoint price is the billed one-image line; 2.1's constraint note says the list rate is the 1K default and the charge scales for 512, 2K and 4K, and Pro's says the 4K list is higher than its default.

Side by side

The shared ratios are 21:9, 16:9, 3:2, 4:3, 5:4, 1:1, 4:5, 3:4, 2:3 and 9:16, plus auto.

Pro and the 4:5 post-step

The docs describe Banana Pro as sending aspect_ratio: "4:5" with a native size of about 928 by 1152 at 1K, and say that an exact 1080 by 1350 comes from a documented post-step through the job's target_pixels. Check the pixel size of what you receive before you upload it to a feed.

Which to send

Use 2.1 when the ratio is one of the four strips, when cost is the constraint, or when you want to draft at 512 first. Use Pro when your own tests show it handles a prompt that 2.1 does not; the catalog gives no other reason to pay about 1.9 times as much at the default tier.

An aspect_ratio of 8:1 is outside Pro's list and will be refused, so a pipeline that fails over from 2.1 to Pro needs to fall back to a listed ratio first.

A fallback chain should be written against the descriptors, not against hard-coded model names. Fetch the endpoint record for each candidate, check that your ratio is in its aspect_ratio values, and move on to the next candidate if it is not. The same check works for resolution and for input_references.