Wan 3.0 eight-second clip: $0.50, $1.00 or $2.00 on Sume

An eight-second Wan 3.0 clip on Sume costs $0.50 at 480p, $1.00 at 720p and $2.00 at 1080p. Wan 3.0 is billed per second by resolution, so the price is the per-second rate times eight.

If you send no resolution, the Wan default is 1080p, which is the most expensive tier. Pin the resolution in the request if you are drafting.

Per-second rates

Sume's Wan 3.0 catalog row records the provider list as $0.05, $0.10 and $0.20 per second (fal page copy of 2026-08-24, subject to change). The billable amount is list times 1.25.

What the vendor says

Alibaba Cloud's Model Studio page for the Wan 3.0 video model lists 480P, 720P and 1080P output, up to 30 seconds, and accepts text, image, video and audio inputs; it prices per second by resolution, from about $0.04 at 480P to $0.20 at 1080P depending on the regional deployment (read 2026-10-07).

Sume's own range is 2 to 30 seconds, which makes Wan 3.0 the only catalog row that accepts a 2-second request. Eight seconds is well inside it.

Request

The body below is an image-to-video job. frame_images with first_frame pins the opening shot and the model animates from it; Wan 3.0 also accepts a last_frame.

{
  "model": "wan-3.0",
  "prompt": "The camera drifts left as the leaves move in the wind",
  "duration": 8,
  "resolution": "720p",
  "aspect_ratio": "16:9",
  "frame_images": [{
    "type": "image_url",
    "image_url": {"url": "https://example.com/garden.png"},
    "frame_type": "first_frame"
  }]
}

When 480p is enough

Eight seconds at 480p is a cheap way to test a prompt before you pay for a final. Because Sume has no seed field, a second run is a new take, not a repeat. Treat the draft as a check on motion and composition, then spend on the keeper.

For a longer view of the same family, the 30-second table in the related post shows the whole range up to 30 seconds.