Wan 3.0 prompt guide

Wan 3.0 Prompt Guide

Six prompt patterns for Alibaba's Wan 3.0, one for each thing the model is known for: long single takes, documents as input, product reveals, multi-reference scenes, data charts and frame-to-frame transitions. Wan 3.0 runs on Alibaba Cloud, not on Imgo — use these prompts with Alibaba's API, or adapt them for Seedance 2.0 on Imgo.

External reference — not available on ImgoLast checked: 2026-10-07

Wan 3.0 at a glance

Model id
wan3.0-video, plus a faster wan3.0-video-prime tier
Status
Public beta since August 6, 2026; callable on Alibaba Cloud Model Studio without an application since August 24, 2026
Length
2 to 30 seconds per request, as one continuous take, at 30 fps
Resolution
480p, 720p or 1080p, with native audio
References
Up to 10 images, 5 video clips and 5 audio clips, plus one document or one public web link
Documents
Word, Excel, PowerPoint, PDF, plain text and Markdown files, or a public web page
Weights
Closed, API only. The last open-weight Wan flagship is Wan 2.2 (July 2025, Apache-2.0)

Availability

Imgo does not run Wan 3.0: there is no Wan 3.0 model in Imgo's generator and no Imgo credit price for it. To run it, use Alibaba Cloud Model Studio or Qwen Cloud; model availability varies by region. To generate on Imgo today, use Seedance 2.0 — the adaptation notes below show how.

Credits and pricing

Not priced on Imgo. Official pricing: Alibaba Cloud bills Wan 3.0 per generated second in three resolution tiers. Its launch post lists $0.05 per second at 480p, $0.10 at 720p and $0.20 at 1080p, so a 30-second 1080p take costs about $6. Check Alibaba Cloud's current rates before you budget.

Six prompts to start from

Replace every bracketed variable. Each prompt says what to attach and what to check in the result.

A 30-second single take

One continuous 30-second take, no cuts. [subject] walks through [location] from [start point] to [end point]. The camera follows at shoulder height on a steady gimbal and slowly arcs to the left at the halfway mark. Ambient sound of [environment], footsteps audible. Natural [time of day] light. Keep the face, clothing and walking pace of [subject] the same for the whole take.

Attach: Optional: one image of the subject.

Check: Compare the subject at second 25 with second 2, and confirm the camera arcs once, not twice.

A document becomes an explainer

Turn the attached document into a 30-second explainer video. Open on the document's title as on-screen text, then show its three key points in order, each as a short visual scene with one line of narration. Calm, confident voiceover in [language]. End on [call to action]. Use only facts that appear in the document.

Attach: One PDF, Word or PowerPoint file — one document or one public link per request.

Check: Check every number, name and date against the source document; nothing should be invented.

Product reveal

A 15-second product reveal of [product] on [surface]. Open in darkness; a single rim light sweeps across to reveal the silhouette, then a soft key light fills in as the camera dollies in to a close-up of [detail]. A subtle [sound] as the light moves. Keep the label text, logo position and colors exactly as in the reference images. No extra products, no captions.

Attach: One to three owned packshots of the product, label legible.

Check: Label spelling, logo placement and color against the packshots.

Two characters, one scene

Use the first reference image as [character A] and the second as the location. [character A] sits at a café table and looks up as [character B] arrives. They exchange two lines: [character A] says "[line 1]" and [character B] answers "[line 2]". Medium two-shot, then a slow push-in on the reaction of [character A]. Quiet café ambience, no music.

Attach: A character image and a location image; optionally a voice sample as an audio reference.

Check: Lip sync on both lines, and the face of the character after the push-in.

A spreadsheet becomes a chart

Using the attached spreadsheet, create a 20-second motion-graphics video that shows [metric] by [period] as an animated bar chart. The bars grow in order, the highest value is highlighted in [brand color], and a caption states the takeaway: "[takeaway]". Clean flat style on a white background, a light whoosh as each bar grows, no music.

Attach: One Excel file with a clearly labelled table.

Check: Every bar value against the spreadsheet, and whether the axis labels are readable.

First frame to last frame

Start exactly on the first-frame image and end exactly on the last-frame image. Over 10 seconds, transform [scene A] into [scene B]: [connecting motion]. One continuous camera move, no cuts, with the light coming from the same direction throughout.

Attach: A first-frame image and a last-frame image with the same aspect ratio.

Check: The final frame should land exactly, with no jump in the last half second.

Running these on Imgo instead

Imgo does not run Wan 3.0, but Seedance 2.0 on Imgo reads the same structure — subject, action, camera, sound — with image, video and audio references. Two changes: Seedance 2.0 has no document input, so paste the document's key points into the prompt; and its clips run up to 15 seconds, so split a 30-second take into two shots that share a reference image.

Seedance 2.0 on Imgo — open the video generator →

Frequently asked questions

Can I use Wan 3.0 on Imgo?
No. Wan 3.0 is not integrated into Imgo and has no Imgo credit price; this page is an external reference, last checked on 2026-10-07. To generate on Imgo, use Seedance 2.0 or pick another model from the live models list.
Is Wan 3.0 open source?
No. Wan 3.0 is API-only and its weights are closed. The most recent open-weight Wan flagship is Wan 2.2, released in July 2025 under Apache-2.0.
How long can a Wan 3.0 video be?
Each request returns 2 to 30 seconds at 30 fps as one continuous take, at 480p, 720p or 1080p with native audio.
Which files can Wan 3.0 read?
Word, Excel, PowerPoint, PDF, plain text and Markdown files, or one public web page that needs no login — one document or one link per request, next to up to 10 images, 5 video clips and 5 audio clips.

Sources

Keep going