Wan 3.0: A Guide to 30-Second AI Video
Alibaba released Wan 3.0 in public beta on August 6, 2026 and made it generally available on August 24. Its headline change is length: up to 30 seconds of video in a single generation, double the 15-second limit of the previous version, with dialogue, music and sound effects generated alongside the picture. Wan 3.0 is API-only, served through Alibaba Cloud Model Studio. This guide covers the confirmed specs, what it costs, and how to plan a 30-second take. Last checked: 2026-10-08.
Specs at a glance
| Spec | Wan 3.0 (wan3.0-video) |
|---|---|
| Clip length | 2 to 30 seconds (default 5), or a "smart duration" the model picks |
| Frame rate | 30 fps |
| Resolution | 480p, 720p or 1080p (1080p by default) |
| Aspect ratios | Adaptive, 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 |
| References | Up to 10 images, 5 video clips (15 s total) and 5 audio clips (15 s total), plus one document (up to 100 MB and 50 pages) or one web link |
| Audio | Native dialogue, background music and sound effects; turning audio off does not change the price |
| Price | $0.05 / $0.10 / $0.20 per second at 480p / 720p / 1080p (international list price) |
The document input is new for the series: a script, brief or slide deck can be passed in directly instead of being pasted into the prompt.
Standard or Prime?
Wan 3.0 comes in two tiers. wan3.0-video-prime has the same capabilities as the standard model, with what Alibaba calls significantly faster end-to-end generation, at $0.068 / $0.14 / $0.28 per second. A full 30-second clip at 1080p costs $6.00 on the standard tier and $8.40 on Prime. Prime pays off when you are waiting on a render; for overnight batches, the standard tier does the same job for less.
Is Wan 3.0 open source?
No. Unlike earlier Wan releases, Wan 3.0 has no published weights and can only be used through the API. The most recent open-weight flagship in the family is still Wan 2.2, released in July 2025 under the Apache-2.0 licence.
How to plan a 30-second single take
- Write a beat sheet before the prompt. 30 seconds at 30 fps is 900 frames. Split it into three to five beats of six to ten seconds, each with one clear action.
- Keep one camera move per beat. In a single take, the camera never cuts; describe how it travels from one beat to the next instead.
- Anchor identity with references. Pass the character, product or location as reference images, so that the subject in the last second still matches the first.
- Budget the dialogue. Natural speech runs at about two to three words per second, so leave space between lines rather than filling all 30 seconds.
- Draft cheap, finish sharp. A 30-second draft costs $1.50 at 480p against $6.00 at 1080p. Settle timing and blocking at low resolution first.
- Extend instead of restarting. Wan 3.0 can continue an existing clip forwards or backwards, as long as input and output together stay within 30 seconds.
Prompt structure
Alibaba's own guide writes long prompts as timestamped segments, each with a camera line and a scene line. Put the global style, lighting and sound first, then the timeline:
Style: warm documentary realism, golden-hour light, 16:9.
Audio: quiet street ambience, soft acoustic guitar under the dialogue.
(0:00 - 0:08) Camera: wide shot, slow push-in. Scene: a cyclist rides into an empty market square.
(0:08 - 0:20) Camera: continues to a medium shot, handheld follow. Scene: she stops at a flower stall. Dialogue (vendor): "You're early today."
(0:20 - 0:30) Camera: slow orbit to a close-up. Scene: she smiles and picks up a bunch of sunflowers.Prompts can run to 20,000 characters, so there is room for detail, but each segment should still describe one action. More ready-made examples are in our Wan 3.0 prompt collection, and the Wan 3.0 overview summarises the model.
Is Wan 3.0 on Imgo?
Not yet. Wan 3.0 is not integrated into Imgo and has no Imgo credit price. What you can use on Imgo today is Seedance 2.0 in the video generator, with clips of up to 15 seconds, image, video and audio references, and native audio. The live model list shows everything you can generate right now, and our Wan 3.0 alternatives page covers the other options. For a side-by-side with the other 30-second model in this class, see Wan 3.0 vs Seedance 2.5.
Sources
Share
Related Posts
No image
Kling 4.0 Flash opened in early access on September 28 for a limited group — reported as Ultra yearly subscribers — with 3–20 second 720p clips. Here is how to get in, what it costs, how to write the 8,000-token prompts and keyframe plans the full model will accept, and what to generate meanwhile.
No image
Wan 3.0 accepts a document, a web link, images, videos and audio as inputs. This operator's guide covers what each input is for, how to structure a 30-second-take prompt, per-second budget examples at 480p to 1080p on both tiers, and what commonly goes wrong.
No image
Kling 4.0 Flash has been in early access since September 28, 2026; the full model and its API are pending, with an October 2026 target. Here is what each tier does, which headline specs are only announced, and what to use in the meantime.