104 newsletter headers in 4:1: $7.80 at 0.5K on Nano Banana 2.1

A year of twice-weekly newsletters needs 104 wide headers: $7.80 at 0.5K or $10.40 at 1K on Nano Banana 2.1 on Sume, the GPT 2.5 3:1 cap aside.

4 min readSume
All posts

The short answer

Two newsletter issues a week for a year need 104 headers, which cost $7.80 on Sume with Nano Banana 2.1 at 0.5K (104 x $0.075) and $10.40 at 1K (104 x $0.10). A 4:1 banner is an extreme ratio, and the Sume docs say GPT Image 2.5 limits the aspect ratio to 3:1 at most.

That makes Nano Banana 2.1 the model on Sume for a 4:1 header, and the launch this week does not change that.

The arithmetic

The GPT limit comes from the documented custom-size rules: both edges must be multiples of 16, the maximum edge is 3840, the ratio is at most 3:1, and the image must have 655,360 to 8,294,400 pixels. A 4:1 image breaks the third rule.

The table prices 104 headers on the models that can produce them, and shows a 3:1 GPT Image 2.5 option for a header you can crop to 3:1.

104 newsletter headers, as of 2026-10-08 (totals rounded to the cent; GPT row assumes a 3:1 header; Google row is a reference)
OptionCount x unit priceTotal
Nano Banana 2.1 at 0.5K104 x $0.075$7.80
Nano Banana 2.1 at 1K104 x $0.10$10.40
Nano Banana 2.1 at 2K104 x $0.15$15.60
GPT Image 2.5 medium (2K)104 x $0.055625$5.79
Google direct, 1K standard (not on Sume)104 x $0.0336$3.49

How to run it

A header does not need a new image every time. Generate 13 headers once a quarter, which is $1.30 at 1K, and rotate them.

Read the aspect_ratio values the model lists in the catalog before you send a ratio. A request with a value the model does not list is rejected.

Retries and cost per issue

Some images will be rejected by eye, and rejected images are billed like any other, so budget a margin. The bullets add 20 percent to each Sume total in the table above (the totals multiplied by 1.2) and divide by the number of issues to give a per-issue cost.

Treat the per-issue figure as the number to compare across models, because it does not change when the count changes. Re-read the public catalog before a large run, since prices can change and the table is a snapshot from 2026-10-08.

  • Nano Banana 2.1 at 0.5K: $7.80 becomes $9.36 with 20 percent extra images, which is $0.09 per issue (a 104-issue job).
  • Nano Banana 2.1 at 1K: $10.40 becomes $12.48 with 20 percent extra images, which is $0.12 per issue (a 104-issue job).
  • Nano Banana 2.1 at 2K: $15.60 becomes $18.72 with 20 percent extra images, which is $0.18 per issue (a 104-issue job).
  • GPT Image 2.5 medium (2K): $5.79 becomes $6.95 with 20 percent extra images, which is $0.07 per issue (a 104-issue job).

What the docs say to check

Read GET /v1/images/models before you pin a tier or ratio. A model accepts only the values its descriptors list, and the descriptors are the definitive source for what a call may send.

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.

Email clients scale wide images unevenly, so test a header on a phone before you commit to a year of them.

The numbers above are list prices read on 2026-10-08, not measurements. Sume figures are the catalog prices that already include its margin, Google figures are from its own pricing page, and quality or speed differences are not covered. Re-read both pages before you commit a large budget, because either side can change its rates.

Sources

Related posts

More in Use cases

All Use cases posts

Written by Sume