A 12s 1080p clip from two 6s Omni clips: 236 cents with the join
Gemini Omni Flash 1.1 stops at 10 seconds, so a 12-second 1080p shot is two 6-second clips at 113 cents each plus a Timeline join: 236 cents on Sume.

Gemini Omni Flash 1.1 on Sume accepts 3 to 10 seconds, so a 12-second 1080p shot cannot be one Omni clip. Two 6-second 1080p clips cost 2 x 113 = 226 cents, a Timeline 1.0 join adds 10 cents, and the finished 12 seconds costs 236 cents ($2.36).
That is cheaper than seedance-2-mini at 511 cents for one 12-second clip and cheaper than wan-3.0 at 300, but it is more than kling-3 with audio off at 168. The saving is real only if you want Omni's look; there are cheaper single-clip routes.
The cost of each route to 12 seconds
Omni's 1080p tier is $0.15 per second list, so a 6-second clip is $0.90 list and $1.125 billable, which rounds up to 113 cents. Timeline 1.0 is $0.10 for each started output minute, and 12 seconds is one started minute.
| Route | Calls | Cost | Notes |
|---|---|---|---|
| Omni 6 s + Omni 6 s + Timeline join | 3 | $2.36 | Two generations, one cut point |
| Omni 10 s alone | 1 | $1.88 | 10 seconds, not 12 |
| wan-3.0, one 12 s clip | 1 | $3.00 | No join |
| kling-3, 12 s, audio off | 1 | $1.68 | No join |
| seedance-2-mini, one 12 s clip | 1 | $5.11 | No join |
What a join means for the shot
Two generations are two separate takes. Even if the second prompt describes the same scene, the model does not carry the first clip's subject into it. If the shot is a single continuous move, you will see the seam; if it is two beats in the same room, a cut hides it.
Omni accepts reference images and short reference videos, and a clip that starts from the last frame of the first one keeps the subject closer. The Video Router docs describe the image_url field for a first frame, so you can extract the final frame of clip one and feed it to clip two.
- Write the two prompts as consecutive beats, not as one prompt cut in half.
- Use the same resolution and aspect ratio for both clips, or the join will rescale one of them.
- Keep the Timeline call out of the loop until you have checked both clips.
The arithmetic for one shot
Omni at 1080p is $0.15 per second list. Six seconds is $0.90, and times 1.25 it is $1.125, which rounds up to 113 cents. Two clips are 226 cents. Timeline 1.0 charges $0.10 for each started output minute, so the 12-second join is 10 cents. The sum is 236 cents, or $2.36. A single 10-second Omni clip is 188 cents, so the extra two seconds and the join cost 48 cents more than stopping at 10.
When to skip the join
For 250 shots, the Omni join route is $590.00 and the single-clip Kling route is $420.00. The 12-second shots that are really two beats are good Omni jobs. The ones that must be one continuous move should go to wan-3.0, kling-3 or a Seedance model.
The Video generation docs list Omni's 3 to 10 second range and the other models' ranges; confirm the range on GET /v1/videos/models for the exact model you pin.
Sources
Related posts
More in Developers
- 1,200 SKUs as 12 bulk queues of 100: keys, concurrency, replays
Format bulk runs take 1 to 100 items, so 1,200 SKUs are 12 queues. Use 12 distinct Idempotency-Keys: a replayed key returns 202 with the old queue.
- 200 video slots in a 180 s Short: 0.9 s each on Sume Timeline
Timeline 1.0 takes up to 200 video slots and 1,800 s of audio. In a 180 s Short, 200 slots average 0.9 s, the render bills $0.30, and fades cap at 0.45 s.
- 250 prompts at 12s 1080p after Sora: which Sume models reach it
Four of the six candidate models make a 12-second 1080p clip on Sume, from $420.00 to $2,042.50 for 250 prompts. minimax-h3 and Omni Flash cannot.
- 48 or 50 fps for YouTube: Sume output.fps accepts 24, 25, 30, 60
YouTube lists 24, 25, 30, 48, 50 and 60 fps as common rates. Sume Timeline's output.fps takes 24, 25, 30 or 60, so omit it for 48 or 50 sources. Why.
Written by Sume