Wan 3.0 for 29 seconds: $1.81 at 480p, $7.25 at 1080p on Sume

Wan 3.0 takes 2 to 30 seconds per job. A 29-second clip costs $1.81 at 480p, $3.63 at 720p and $7.25 at 1080p on Sume. Arithmetic and limits inside.

5 min readSume
All posts

A 29-second Wan 3.0 clip on Sume is $1.81 at 480p, $3.63 at 720p and $7.25 at 1080p. Wan 3.0 accepts 2 to 30 seconds in one job, which makes it, with Seedance 2.5, one of the two rows that reaches a half-minute shot. Wan is also the row with the lowest per-second price at 1080p in the Sume list: $0.25 against $1.42155 for Seedance 2.5.

Price by resolution

Rates are linear in duration, so you can scale any figure in the table.

Wan 3.0 on Sume, as of 2026-10-09 (default options, list x 1.25)
ResolutionSume rate per second29 s arithmetic29 s price2 s minimum30 s maximum
480p$0.062529 x $0.0625 = $1.8125$1.81$0.13$1.88
720p$0.12529 x $0.125 = $3.625$3.63$0.25$3.75
1080p$0.2529 x $0.25 = $7.25$7.25$0.50$7.50

Inputs Wan 3.0 accepts on a long clip

Wan 3.0 supports text-to-video, image-to-video with first and last frames, and reference-to-video with images, video and audio references. The Video Router constraints list caps on those references: up to 10 reference images, up to 5 reference videos (15 seconds in total, at least 16 fps) and up to 5 reference audio clips (15 seconds in total).

Aspect ratios on this row are auto, adaptive, 16:9, 4:3, 1:1, 3:4 and 9:16. 21:9 is not on the list.

  • Duration range: 2 to 30 seconds, whole seconds.
  • Resolutions: 480p, 720p, 1080p.
  • Send Idempotency-Key when you retry a long job so a replay returns the original job instead of a second charge.

When 29 seconds is the wrong plan

A single 29-second generation is the most expensive way to find out your prompt is off. A common pattern is to draft at 480p for $1.81, review the motion, and only then run the 1080p version for $7.25. The draft costs one quarter of the final, because 480p is $0.0625 per second against $0.25 at 1080p.

Poll with a moderate interval; the docs suggest about 30 seconds, and long clips can take several minutes.

Checking the result

A 30-second ceiling means 29 seconds leaves one second of headroom if you plan to trim the head or tail. Download from unsigned_urls[0], inspect it with video inspect for duration, resolution and frame rate, and only then send it into a timeline.

Video inspect probes a Sume-hosted clip and returns sampled stills, so it is the cheapest way to confirm the clip really is the length and size you ordered before you build a longer edit around it.

Sources

Related posts

More in Pricing

All Pricing posts

Written by Sume