Grok Imagine on Sume: 3 cents, one image per call, 9:20 phone ratios
Grok Imagine bills 3 cents per image on Sume, returns one image per call, lists 13 ratios including 9:20 and 19.5:9 phone shapes, and supports edits.

Grok Imagine (grok-image) bills 3 cents per image on Sume, returns one image per call, accepts edits, and is the only catalog model that lists the phone-screen ratios 9:20, 9:19.5 and 19.5:9. A hundred images cost $3, but that is 100 separate calls, because the catalog caps it at one image.
What the catalog lists
List price is $0.02, times 1.25 is $0.025, rounded up to 3 cents. The row lists 13 ratios and png, jpeg and webp output, and no quality or resolution tiers.
| Field | Value |
|---|---|
| Billed per image | 3 cents |
| 100 / 500 / 1,000 images | $3 / $15 / $30 |
| 10,000 images | $300 (10,000 calls) |
| Images per call | 1 |
| Edits | yes |
| Phone ratios | 9:20, 9:19.5, 19.5:9, 20:9 |
| Other ratios | 2:1, 1:2, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 1:1 |
| 4:5 | not listed |
Using the phone ratios
A wallpaper or a full-screen app background fits a modern tall phone better at 9:20 than at 9:16. Grok Imagine is the row to use when you want that shape from the model itself rather than a crop. The cost of the extra care is the call count: with n capped at 1, a set of 12 wallpapers is 12 calls, so use async or webhook mode in a loop and let Sume's job envelope carry the waiting.
Because the cap is one, price comparisons with n=4 rows are not apples to apples in time, only in dollars. Qwen Image bills the same 3 cents and returns 4 per call, and it lists 4:5 where Grok does not.
- Phone wallpaper at 9:20: Grok Imagine.
- Instagram 4:5: use another row.
- High volume with n=4: Qwen Image or Imagen 4 Fast at the same 3 cents.
Edits
Grok Imagine accepts references, which makes it the cheapest edit-capable row tied with Qwen Image. Send public HTTPS URLs and test one before you queue a long run.
Queueing one-image calls
Because each call returns one image, queue calls in parallel instead of in a row. Each completes inside the 30-second sync wait for most prompts, and any that run longer return a 202 job you can poll. Cap your parallelism, and if you get a 429, back off and retry with a delay rather than in a tight loop.
For a 30-wallpaper pack at 9:20, plan 30 calls and 90 cents. For a 300-wallpaper pack, 300 calls and $9. If you also want 4:5 posters in the same campaign, send those to Qwen Image at the same 3 cents with n=4.
Sources
Related posts
More in Models
- H3 on vLLM-Omni: 87 s on 4 B300 for one clip, and the cost
MiniMax says an 8.7 s H3 clip takes about 87 s on 4 B300 GPUs. Turn that into GPU-seconds, a break-even rate, and compare with a 9 s Sume job.
- Higgsfield Soul on Sume: half a cent per image, batches of 1 or 4
higgsfield-soul costs $0.005 at 720p and $0.0075 at 1080p on Sume. It takes batch sizes 1 or 4, no references. Price table and a four-image call.
- Higgsfield Soul on Sume: 1 cent an image, n of 1 or 4, $100 per 10,000
Higgsfield Soul is the cheapest Sume image model at 1 cent. It is text-only, takes n of 1 or 4 and 720p or 1080p, so 10,000 images is 2,500 calls and $100.
- High to max on 100 hero images: GPT Image 2.5 adds $20, xhigh adds $5
Upgrade 100 GPT Image 2.5 hero images from high to xhigh or max: the extra cost on Sume is $5 or $20 at 1024-class size. The arithmetic is shown.
Written by Sume