Video
Wan 3.0 in Twelver: Alibaba’s video model for 30-second clips with sound
We’ve moved Wan to its third generation. A clip can now run to thirty seconds, sound is generated with the picture, and you can hand it images, video and a voice sample at once.
What’s new
- Clips up to 30 seconds instead of 15 — with sound, dialogue and music generated together with the picture.
- Timestamped storyboards: several shots from one request.
- Up to ten images, five videos and five audio clips as references — characters, products, camera moves, a voice’s timbre.
- Editing existing footage — remove, add or replace an object, restyle it — and extending a clip.
- A new 480p tier — drafts at half the price of 720p.
At a glance
| Detail | Value |
|---|---|
| Model | Wan 3.0 (Alibaba) |
| Replaces | Wan 2.7 |
| Clip length | 2–30 seconds |
| Resolution | 480p, 720p, 1080p |
| Aspect ratios | 16:9, 9:16, 1:1, 4:3, 3:4, 21:9 |
| Sound | Dialogue, effects and music — with the picture |
| References | Up to 10 images, 5 videos, 5 audio clips |
| Where in Twelver | Chat · MCP server |
We shot it ourselves — here is the result
Both clips below were made by Wan 3.0 in Twelver from a single request, with no editing. The first is three shots set by timestamps, with the sound generated alongside the picture.
The second is a product-card clip from one photo: the bottle and the lettering on its label stay as they are in the picture.
Three more examples — animating a frame, a clay-style edit and extending a clip — are in the Wan chapter.
Frequently asked questions
What’s new in Wan 3.0?
Mainly length and sound: clips up to 30 seconds instead of 15, with dialogue, sound and music generated together with the picture. Plus timestamped storyboards, video and voice as references, extending an existing clip, and 480p for cheap drafts. The full breakdown is in the Wan chapter.
How do I try Wan 3.0?
Open the Wan chapter — it has a window with the model itself, and your first try needs no sign-up. You can carry on in Twelver chat.
How much does a Wan 3.0 clip cost?
It is billed per second and depends on resolution: 480p is half the price of 720p, 1080p twice as much. Editing and extending also bill the length of the source video. Current prices are on the video generation page.
Can I call Wan 3.0 from Claude or Cursor?
Yes, over MCP: video generation is available over MCP in Claude, Cursor and other clients — see the MCP page.
Try it
Open in Twelver chatRate your experience