What n: 4 costs on each Sume image model: cost x n, billed table
Sume's image docs say you pay cost_usd x n. Four images cost 4x the per-image rate: $0.10 on Qwen Image, $0.40 on Nano Banana 2.1, $0.75 on Pro. Full table.

On Sume's Images API you pay the per-image price times n. The docs say the endpoint pricing lines already include the margin, so you pay cost_usd x n. Four images cost four times one: $0.10 on Qwen Image, $0.40 on Nano Banana 2.1 at 1K, $0.75 on Nano Banana Pro at 1K.
Billed cost of n: 4
| Row | Per image | n: 4 |
|---|---|---|
| Higgsfield Soul (720p) | $0.005 | $0.02 |
| Imagen 4 Fast | $0.025 | $0.10 |
| Qwen Image | $0.025 | $0.10 |
| Seedream 4.0 | $0.0325 | $0.13 |
| Flux 2 Pro | $0.0375 | $0.15 |
| Seedream 4.5 | $0.05 | $0.20 |
| Ideogram 4.5 (medium) | $0.075 | $0.30 |
| Nano Banana 2.1 (1K) | $0.10 | $0.40 |
| Nano Banana Pro (1K) | $0.1875 | $0.75 |
Caps that change the picture
The request cap for n is 10, but each model has its own ceiling, which the catalog publishes as the n range descriptor. Most rows list a maximum of 4. Grok Imagine lists 1, so four stills mean four calls at $0.025 each. Higgsfield Soul takes 1 or 4. Asking for more than a model allows is rejected, not trimmed, so read the descriptor first.
Failures and refunds
Billing is all or nothing. A completed generation is charged in full, a failed or cancelled one is not charged, and a client that disconnects early is treated as a failed generation. If you ask for four and the call fails, you pay nothing for it.
How to budget a batch
- Multiply the billed per-image price by images wanted, not by calls.
- Use
nto reduce round trips; the price does not change versus single calls. - Estimate from
GET /v1/images/models/{id}/endpoints, then confirm withusage.coston the first response. - For ChatGPT Image 2.5, price depends on quality and size, so use the estimator values instead of the catalog display rate.
Per-call overhead and speed
Because the price is linear in n, the decision to batch is about time and complexity, not money. One call with n: 4 is one request to track and one job to poll; four calls are four jobs, four idempotency keys and four chances for a transient error. On the other hand, a four-image call takes longer to finish, so it is likelier to pass the 30-second sync wait and return 202.
A good default for a UI is n: 1 in sync mode for instant feedback and n: 4 in async mode when the user asks for options. For automation, prefer n: 4 with mode: "async" and one idempotency key per call. The total stays the same and your code has fewer jobs to follow.
Example
A set of 12 thumbnail options as three calls with n: 4 on Nano Banana 2.1 costs 3 x $0.40 = $1.20, the same as twelve single calls. Slow settings (4K, high quality, large n) may return 202, so branch on the status code, as the Image API page explains.
Sources
Related posts
More in Pricing
- What Sume's 1.25 margin adds per 100 clips on Omni and Wan
Sume bills provider list times 1.25 on video-router models. 100 Omni 720p clips of 8 s carry $20 of margin; 100 Wan 1080p clips of 30 s carry $150.
- Sume yearly plans per job slot: Pro $9, Startup $13.13, Scale $17
On yearly billing Sume Pro is $36 a month for 4 job slots ($9 each), Startup $105 for 8 ($13.13) and Scale $340 for 20 ($17).
- Sume yearly vs monthly plan price: the 10, 12.5 and 15 percent math
Yearly stickers are $432, $1,260 and $4,080 against $480, $1,440 and $4,800 monthly: 10, 12.5 and 15 percent off, or $36, $105 and $340 a month.
- YouTube ad length by format, and what each costs to render in Timeline
Bumper 6 s, non-skippable 15 to 60 s, Shorts under 60 s, skippable with no max. Google's lengths mapped to Sume Timeline billable minutes and cost.
Written by Sume