20 s of Wan 3.0: one $2.50 job or two 10 s clips plus Timeline

One 20 s Wan 3.0 job at 720p costs $2.50 on Sume. Two 10 s jobs ($2.50) plus a Timeline render ($0.10) cost $2.60. When the extra dime buys you something.

5 min readSume
All posts

One 20-second Wan 3.0 job at 720p costs $2.50 on Sume. Two 10-second jobs cost $2.50, and joining them on Timeline 1.0 adds $0.10 for a render of up to one minute, so $2.60. The 10 cents is the price of control: two clips can be cut, reordered, and re-rolled separately. A single job has no join and no second render.

The arithmetic

Wan 3.0 is priced per second by resolution. At 720p the catalog list is $0.10 per second, billed at list x 1.25. Timeline 1.0 charges $0.10 per output minute, rounded up.

Billable price of 20 s at 720p, Sume as of 2026-10-08
RouteCalculationPrice
One 20 s job0.10 x 20 x 1.25$2.50
Two 10 s jobs2 x (0.10 x 10 x 1.25)$2.50
Timeline 1.0 join$0.10 x 1 output minute$0.10
Two jobs plus Timeline2.50 + 0.10$2.60

When two clips are the better buy

  • One of the two shots fails. Re-rolling a 10 s clip costs $1.25. Re-rolling the whole 20 s job costs $2.50.
  • You want a cut: a hard cut at 10 s that looks intentional is easier to get from two prompts than from one long one.
  • You need different first frames, because each job takes its own first and last frame.
  • You already render on Timeline for captions, audio or an end card, so the join is free of extra steps.

When one job is the better buy

Choose one job when the shot is a single continuous action, such as a slow walk or a camera move, where a cut would be visible. Wan 3.0 accepts 2 to 30 seconds, so 20 s is inside its range. You also skip the risk of a visible seam, which depends on how the second clip starts. The catalog allows a last frame to seed the next first frame through frame_images, but the second clip is a new generation, so a seam can still appear.

How to run it

Submit pinned jobs through Video generation with model: "wan-3.0", resolution: "720p" and the duration you want, and poll the returned polling_url. Use Idempotency-Key so a retry returns the original job. For the join, read the Timeline 1.0 page; its bill is $0.10 per ceil output minute, so a 20 s output costs the same as a 59 s one.

A last check before you batch this: the $0.10 join is a flat charge per output minute, so it does not grow with clip price. If you pick a dearer model or resolution, re-run the arithmetic with the grid price for that row; the join stays at ten cents while each re-roll costs more, which makes separate clips more attractive as clips get dearer.

Check it before you run it

Every figure above is a catalog list price times 1.25, rounded up to the cent, as of 2026-10-08. Catalogs change, so before a large batch, read the current model entry in the docs and recompute the one line that matters for your case. Write the arithmetic next to the job in your own notes: list rate, seconds or characters, multiplier, rounding. If the result differs from the wallet charge by more than a cent, the catalog entry has changed, and the docs page is the place to find out why.

Run one small job first. Submit a single request with the pinned model id and the settings in the tables, poll the returned polling_url until it finishes, and compare the charge with your estimate. Then scale up. Using a pinned id for the test matters, because sume/auto never names the family, so you cannot tie its charge to the row you priced.

Sources

Related posts

More in Comparisons

All Comparisons posts

Written by Sume