Six 5-second reference clips: $2.28 on H3 to $17.34 on Seedance 2.5

Cost of six 5-second 720p reference-to-video clips on Sume by model: MiniMax H3 $2.28 at 768p, Wan 3.0 $3.78, Seedance 2.0 Fast $9.12, Seedance 2.5 $17.34.

5 min readSume
All posts

Six 5-second clips cost $2.28 on MiniMax H3 at 768p, $3.00 on H3 Max, $3.78 on Wan 3.0 or Gemini Omni Flash 1.1 at 720p, $5.70 on Seedance 2.0 Mini, $9.12 on Seedance 2.0 Fast and $17.34 on Seedance 2.5. All of those models list reference images in the Sume catalog. These are clip prices only; the notes below say what is missing.

The price per clip and for six

Prices are billable cents per clip from the Sume catalog formula: provider list x 1.25, rounded up to a whole cent. Seedance rows are for 9:16 at 720p; Wan, MiniMax and Omni rows do not vary with aspect.

Six 5-second clips with references, by model (read 2026-10-08)
ModelResolutionOne clipSix clipsReference images
MiniMax H3768p38c228c = $2.28up to 9
MiniMax H3 Max768p50c300c = $3.00up to 9
Wan 3.0720p63c378c = $3.78up to 10
Gemini Omni Flash 1.1720p63c378c = $3.78up to 10
Seedance 2.0 Mini720p95c570c = $5.70listed, count not in catalog note
Seedance 2.0 Fast720p152c912c = $9.12listed, count not in catalog note
Seedance 2.5720p289c1734c = $17.34listed, count not in catalog note

Check the arithmetic on one row

Wan 3.0 lists $0.10 per second at 720p. Five seconds is $0.50. With the 1.25 margin that is $0.625, which rounds up to 63 cents. Six clips are 6 x 63 = 378 cents. H3 at 768p lists $0.06 per second: 5 x 0.06 = 0.30, x 1.25 = 0.375, up to 38 cents, and 6 x 38 = 228 cents. Gemini Omni Flash 1.1 lists $0.10 per second at 720p, the same 63 cents.

Two caveats. Omni takes reference videos of 3 seconds or less, takes no audio references, and has a native audio track that is always on. And H3 charges at list for reference images after the fifth, so a six-image set adds one image at $0.08 list.

Which row suits which job

  • Testing a reference set: MiniMax H3 at 38c a clip, 768p native.
  • Fast iteration with audio references: Wan 3.0, which lists up to 5 audios of 15 seconds combined.
  • Short clips up to 10 seconds with native audio: Gemini Omni Flash 1.1, 3 to 10 seconds, 16:9 or 9:16 only.
  • Longer clips: Seedance 2.5 runs 4 to 30 seconds; Wan 3.0 runs 2 to 30 seconds.
  • Final quality where the price is not the limit: the Seedance rows, but check a draft elsewhere first.

What the table leaves out

The totals above leave out the cost of making the reference images, any retries, and the Timeline render that joins the six clips. Six 5-second clips make 30 seconds, so a Timeline render of that length is ceil(30 / 60) = 1 minute, $0.10. Add that for a finished file. If a clip fails, the table does not cover the redo.

Choose the model for the job, not for the lowest line. Cheap rows are good for tests. A test of six clips on H3 for $2.28 tells you whether your reference set holds a character before you spend $17.34 on the same six on Seedance 2.5.

Sources

Related posts

More in Media tools

All Media tools posts

Written by Sume