Still plus clip: how many variants $25 buys with Grok Video 1.5
Grok Imagine Video 1.5 needs a start image. Pair it with a $0.007375 GPT Image 2.5 low still and $25 buys 435 four-second variants; with Nano Banana 2.1, 166.

Grok Imagine Video 1.5 on Sume is image-to-video only, so every clip needs a start image. At a Sume billable $0.0125 per second, a 4-second clip is $0.05. Add a GPT Image 2.5 low still ($0.007375) and one variant costs $0.057375, so $25 buys 435 four-second variants. With a Nano Banana 2.1 1K still ($0.10), one variant costs $0.15 and $25 buys 166.
Prices are Sume billable amounts: provider list times 1.25, rounded up to a micro-dollar. Grok Imagine Video 1.5 lists $0.01 per output second at 480p and 720p, which is $0.0125 billable. Its Video Router entry takes 4 to 15 seconds, and it has no end frame, reference inputs or audio toggle.
What does one clip cost by length?
The rate is flat per second, so cost is exactly seconds times $0.0125. Switching between 480p and 720p does not change it.
| Clip length | Billable per clip | Clips per $20 |
|---|---|---|
| 4 s | $0.05 | 400 |
| 6 s | $0.075 | 266 |
| 8 s | $0.10 | 200 |
| 10 s | $0.125 | 160 |
| 15 s | $0.1875 | 106 |
How does the still change the total?
The still is a separate image job, so a variant is two charges: one image and one clip. The table adds them and divides $25 by the total, rounding the count down.
| Still model | 4 s clip | 6 s clip | 10 s clip |
|---|---|---|---|
| GPT Image 2.5 low ($0.007375) | 435 | 303 | 188 |
| GPT Image 2.5 medium ($0.0165) | 375 | 273 | 176 |
| Nano Banana 2.1 1K ($0.10) | 166 | 142 | 111 |
What does a 20-variant test cost?
A typical creative test is 20 variants at 6 seconds. With GPT Image 2.5 low stills that is 20 times $0.082375, or $1.6475. With GPT Image 2.5 medium stills it is 20 times $0.0915, or $1.83. With Nano Banana 2.1 1K stills it is 20 times $0.175, or $3.50.
The clip is the larger share in the cheap case: $0.075 of the $0.082375 low-still variant. The still is the larger share when you pick Nano Banana 2.1 at 1K, where $0.10 of the $0.175 is the image. Spending more on the still only pays off if the still quality is what you are testing.
Will my wallet count match the table?
The counts above use exact micro-dollar amounts. The Video Router catalog also lists whole-cent amounts rounded up, so if each of your requests were rounded up to a cent, the 4-second low-still figure would be 6 cents per variant and $25 would buy 416 instead of 435. Plan with the lower number if a batch has to fit a hard limit, and read the real amount from the billable field in each job response.
Sources
Related posts
More in Pricing
- Syren's 2,000 launch credits end Nov 9: what 2 minutes costs on Sume
Syren's launch credits cover about one or two 2-minute videos before Nov 9. A 2-minute Sume avatar video is $22.08, $29.40 or $66.00 by tier.
- Ugly Christmas sweater listings: 60 cutouts and 10 heroes, $3.35
Background removal for 60 sweater photos costs $1.35 on Sume at $0.0225 each. Upscaling the 10 best adds $2.00, for $3.35 in total.
- One agent turn's total cost: pass its job id to /v1/usage
Pass an agent turn's job id as job_id on GET /v1/usage and Sume sums the turn's own model row plus every job it commissioned, with turn_job_id on each row.
- Ten sub-cent usage rows: why cents fields never match debited_usd
Sume prints three roundings for the same money: credit_amount rounds up, billable_amount_usd_cents rounds half-up, debited_usd rounds once. Sum micros instead.
Written by Sume