Wan 3.0 on Sume: a 2-second minimum clip and a 30-second max, priced

Wan 3.0 (wan-3.0) accepts clips from 2 to 30 seconds, the widest range in the Sume video catalog. A 2-second clip costs $0.125 at 480p, $0.25 at 720p and $0.50 at 1080p. A 30-second clip at 1080p costs $7.50.

Wan 3.0 is billed per output second by resolution, which makes the arithmetic simpler than the token-priced Seedance family. The table shows the per-second rates and four common lengths, as of 2026-10-08.

How to read the table

The provider list is $0.05 per second at 480p, $0.10 at 720p and $0.20 at 1080p, from the fal list recorded in the Sume pricing package on 2026-08-24, which the code marks as subject to change. Sume multiplies each rate by 1.25.

Example: 5 seconds at 720p is 5 x $0.10 = $0.50 list, and $0.50 x 1.25 = $0.625. That matches the Sume test fixture for a 5-second 720p Wan 3.0 request, which reserves $0.625.

What a 2-second minimum is good for

Most video models in the catalog start at 4 or 5 seconds: Seedance and Kling at 4, MiniMax H3 and H3 Max at 5, Gemini Omni Flash 1.1 at 3. Wan 3.0 is the only generation model that goes down to 2. That suits looping stingers, logo reveals, reaction cuts and single-gesture clips where a 4-second minimum would force you to trim.

Short clips also make prompt tests cheap. Ten 2-second drafts at 480p cost 10 x 2 x $0.0625 = $1.25.

• Minimum: 2 seconds. Maximum: 30 seconds.
• Resolutions: 480p, 720p, 1080p.
• Aspect ratios: auto, adaptive, 16:9, 4:3, 1:1, 3:4, 9:16 (no 21:9).
• Native audio is on, and the catalog flags no bitrate mode.

Request example

Pin the model, give an integer duration, and name the resolution so the reserve matches the table.

curl -X POST https://api.sume.com/v1/video-router/generate \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: wan-2s-001" \
  -d '{"model":"wan-3.0","prompt":"A coin spins and lands flat","resolution":"480p","duration":2,"aspect_ratio":"1:1","mode":"async"}'

Limits to remember

The Sume docs state that wan-3.0 accepts 2 to 30 seconds and that Wan 3.0 accepts audio and video references. Read capabilities from GET /v1/video-router/models for the envelope your key sees.

Cost of the other sizes in the same family: 1080p is the most expensive tier at twice the 720p rate, and the 480p tier is the cheapest way to test a prompt before a final render.

A reminder on retries: send an Idempotency-Key on every submit. A replay returns the original job, so a network timeout does not double bill you for a 30-second clip. For long clips prefer callback_url (HTTPS only) over a tight polling loop; Sume suggests a moderate interval such as 30 seconds when you do poll.

Sume signs webhook bodies with x-sume-webhook-signature, so verify the header before you trust the payload, and refuse to start your receiver if the signing secret is empty.