Wan 3.0
Wan 3.0 on Scenes AI
Wan 3.0 is an AI video model made by Alibaba, and one of the engines you can generate with on Scenes AI. Up to 30s in one generation · 1080p with native audio · 10 reference images. You pay per second at the rates below, with no subscription — and Scenes AI handles feeding it your character and location references so what it produces stays consistent with the rest of your video.
What it does
The capabilities and limits that matter when you pick Wan 3.0.
- Scenes up to 30 seconds long.
- Aspect ratios: adaptive, 16:9, 4:3, 1:1, 3:4, 9:16.
- Resolutions: 480p, 720p, 1080p.
- Accepts up to 10 reference images — this is what carries your character sheets and location plates into the shot.
- Supports last-frame interpolation, so one scene can morph into the next instead of cutting.
- In-video audio can be switched on or off per scene.
- Billed per second of output, so you pay for exactly the length you use.
Pricing
What Wan 3.0 costs, per second. One credit is one US dollar.
| Resolution | Rate per second |
|---|---|
| 480p | $0.052 |
| 720p | $0.12 |
| 1080p | $0.21 |
A finished video is more than one line item — see the pricing calculator for what a whole video costs with Wan 3.0 selected.
Keep reading
- All supported modelsEvery video, image and music engine available, side by side.
- Pricing and calculatorPrice a whole video, not just one engine's rate.
- How character consistency worksWhat those reference images are actually doing.
- Wan 3.0 vs Seedance 2.5The quality-versus-price trade-off, with real numbers.