Wan 3.0 doubles clip length to 30 seconds: Alibaba ids vs Sume
Alibaba Model Studio lists wan3.0-video and wan3.0-video-prime at 2 to 30 seconds, up from 15 on Wan 2.7. What Sume's wan-3.0 accepts and a 30 s request.

Wan 3.0 clips run 2 to 30 seconds. The Alibaba Cloud Model Studio page lists wan3.0-video and wan3.0-video-prime at 2 to 30 seconds with 480P, 720P, and 1080P, while Wan 2.7 and 2.6 are listed at 2 to 15 seconds. On Sume the catalog id is wan-3.0, also 2 to 30 seconds.
Ids and ranges side by side
The Model Studio ids and the Sume id are different strings for related things. Alibaba lists two Wan 3.0 ids, the second marked prime. Sume's video docs name a single wan-3.0.
| Source | Id | Duration | Resolution |
|---|---|---|---|
| Alibaba Model Studio | wan3.0-video | 2 to 30 s | 480P, 720P, 1080P |
| Alibaba Model Studio | wan3.0-video-prime | 2 to 30 s | 480P, 720P, 1080P |
| Alibaba Model Studio | Wan 2.7 and 2.6 | 2 to 15 s | not restated for this row |
| Sume docs | wan-3.0 | 2 to 30 s | check supported_resolutions |
What people ask
Search suggestions around Wan 3.0 include "wan 3.0 open source" and "wan 3.0 vs seedance 2.5". Neither answer is on the Model Studio page read here. The Sume docs do put Seedance 2.5 and Wan 3.0 in the same group: both reach 30 seconds, while every other catalog model other than a few special routes tops out at 15.
A 30 second request
A single 30 second request needs only the model, a prompt, and a duration. Each model advertises its accepted resolutions, so set resolution from the live list rather than from a vendor page. The video docs describe the polling flow after submit.
curl -X POST https://api.sume.com/v1/videos \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: wan-30s-001" \
-d '{
"model": "wan-3.0",
"prompt": "One continuous shot: a ceramic cup on a wooden table as morning light moves across it",
"duration": 30,
"aspect_ratio": "16:9"
}'When 30 seconds in one call is worth it
Long single shots suit continuous motion: a walkthrough, a slow product turn, a landscape. For multi-scene stories, separate clips joined on a timeline give you more control over each scene.
Cost also matters. A 30 second clip that misses the brief wastes more than a 6 second one that does. Test an idea at 6 seconds first, keep the prompt that works, and only then ask for the long version.
- Continuous motion: one 30 second request.
- Multiple scenes: several shorter clips.
- Unproven prompt: test at 6 seconds first.
- Check
supported_durationsbefore submitting.
Test short, then go long
Thirty seconds in one shot costs more per attempt than six seconds, so test the idea at a short length first, then scale. The related post on single take versus stitched clips covers when to split.
Sources
Related posts
More in Models
- What is FLUX 3? The family map: image, video, audio, action
BFL's docs describe FLUX 3 as one family covering image, video with synchronized audio, audio and action. Which pieces have open weights and which do not.
- An OpenRouter-compatible video API: sume/auto or a pinned model
Sume's POST /v1/videos follows OpenRouter's video generation API field for field. Let sume/auto pick the model, or pin a catalog id like seedance-2.5.
- Image generation API with reference images: POST /v1/images
Send a prompt plus public HTTPS reference images to Sume's POST /v1/images. Pin a catalog model or send sume/auto; the catalog lists each model's limits.
- Video 1.0 and Image 1.0 are retiring soon: move to sume/auto
Sume Video 1.0 and Image 1.0 are retiring soon and already run as aliases for the Auto path. New integrations call /v1/videos or /v1/images with sume/auto.
Written by Sume