Wan 3.0 vs Seedance 2.0: which model to call when
Wan 3.0 and Seedance 2.0 are both on Sume's video API. Wan runs 2 to 30 s and bills per second; Seedance 2.0 runs 4 to 15 s with a 21:9 frame. Pick by the job.

On Sume, choose wan-3.0 when the clip must be shorter than 4 seconds or longer than 15, up to 30, and choose seedance-2 when you want a 21:9 frame or a price that follows frame size. Both take a prompt, first and last frames, and image, video and audio references, so the choice is about length, frame shape and how the job is metered, not about what inputs are allowed.
This post ranks nothing. It lists what each id accepts, from the Video generation docs and the catalog read on 2026-09-29, and one per-second price from fal's Wan 3 page.
How do the two models differ on Sume?
| Item | `wan-3.0` | `seedance-2` |
|---|---|---|
| Clip length | 2 to 30 s | 4 to 15 s |
| Resolutions | 480p, 720p, 1080p | 480p, 720p, 1080p |
| Aspect ratios | 16:9, 4:3, 1:1, 3:4, 9:16 | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 |
| References | Images (up to 10), videos (up to 5), audio (up to 5) | Image, video and audio |
| Meter | Per second, by resolution | Per 1,000 video tokens |
What does a 10-second clip cost on each?
Reserved on submit at 720p 16:9, before the agent fee: $1.25 for wan-3.0 and $3.78 for seedance-2. Wan's amount depends on seconds and resolution only; Seedance counts tokens from width, height and seconds, so a different aspect ratio moves its number.
When should I call Wan 3.0?
- The clip is 2 or 3 seconds, such as a logo sting or a reaction cut.
- One take must run 16 to 30 seconds and you want to avoid stitching.
- You want per-second pricing you can work out by hand before you submit.
- You are passing many references, and want the per-type caps checked at submit.
When should I call Seedance 2.0?
- You need a 21:9 frame, which
wan-3.0does not list. - The clip is 15 seconds or less, and 4 seconds or more.
- You already use
seedance-2prompts and want the same behavior.
Can I let Sume choose?
Yes. Send model: "sume/auto" and Sume picks a family; the response echoes sume/auto and does not say which one ran. Auto create controls default to 720p and 8 seconds, at 3 to 10 seconds in 16:9 or 9:16, so pin wan-3.0 or seedance-2 when you need a length or ratio outside that.
Sources
Related posts
More in Models
- Wan 3.0 web page to video: link input and how to do it on Sume
Wan 3.0 can read a public web link to make a video on Alibaba's API. Sume's wan-3.0 has no link field, so fetch the page yourself and send text and images.
- DALL-E 3 API removed May 12, 2026: GPT Image ids on Sume
OpenAI removed dall-e-3 from its API on May 12, 2026 and points to gpt-image-2. On Sume, send openai/gpt-image-2 to POST /v1/images instead.
- Kazakh speech to text API: Nova-3 and Sume language_code
Deepgram added Kazakh to Nova-3 on Sep 3, 2026. Sume STT 1.0 takes an optional language_code hint, or auto-detect. How to send audio and check the output.
- FLUX video edit API: 720p downscale, and video_url edits on Sume
BFL's edit-a-video endpoint downscales sources above 720p. Sume lists no FLUX video; its edit path is video_url with gemini-omni-flash-1.1 only.
Written by Sume