Latest unified model01
Wan 3.0 Video
Document & webpage-to-video explainers
- Input
- All-modality references in one request
- Duration
- 2–30s
- Resolution
- 480p / 720p / 1080p
Alibaba Cloud · Wan model family hub
Choose the Wan video model by the control your shot needs
Wan 3.0 is Alibaba's all-modality reference video model. Beyond the text, image, and endpoint workflows of earlier Wan versions, it accepts documents and web pages as creative sources and combines image, video, and audio references in a single request — useful when a clip must match existing brand assets, a voice track, or written source material.
Latest connected model
Create with Wan 3.0 Video
The family Hub starts with the latest unified Wan model. Open the version page for detailed input guidance and examples.
Loading...
Model family overview
What is Wan AI video?
Wan 3.0 is Alibaba's all-modality reference video model. Beyond the text, image, and endpoint workflows of earlier Wan versions, it accepts documents and web pages as creative sources and combines image, video, and audio references in a single request — useful when a clip must match existing brand assets, a voice track, or written source material.
Wan version directory
One family Hub, one page per model version
Wan 3.0 is the current unified route. Wan 2.7 remains available for previous workflows, while Wan 2.6 keeps its three connected production routes together. Choose by control and production role, not by URL suffix.
Wan 2.702
Wan 2.7 Video
First/last frames, reference video and audio, multimodal control
- Input
- Text · Image · Endpoints · References
- Duration
- 5 / 10 / 15s
- Resolution
- 720p / 1080p
Three routes, one page03
Wan 2.6
Text-to-video, quality image animation, lower-credit Flash drafts
- Input
- Text · Image Quality · Image Flash
- Duration
- 5 / 10 / 15s
- Resolution
- 720p / 1080p
Choose what must stay fixed
Use an image for identity or shape, endpoints for the final composition, and references for motion or sound.
Write one coherent shot
Subject, action, environment, camera, light, sound, and ending should describe the same production beat.
Review the complete audiovisual result
Check identity, geometry, motion continuity, generated words, sound timing, and the final edit point.
Wan use cases
Match the model workflow to a real deliverable
Product marketing video
Preserve an approved product visual, add deliberate material motion, and finish on a usable hero frame.
Character and performance
Use an image to anchor identity and a short reference to communicate performance timing.
Cinematic previsualization
Turn one story beat into a coherent shot with a clear camera path and ending.
Social video hooks
Test an immediate action and readable payoff before committing to a longer edit.
Wan prompt guide
Reusable prompt patterns for every Wan input path
Cinematic text prompt
[Shot size] of [subject] performing [single action] in [environment]. [Camera move], [lighting], [material/weather motion], [sound cue], finish on [ending composition].
Use this promptControlled image animation
Animate the supplied image while preserving [identity/product geometry/material]. Add [one action] and [one camera move]. Natural motion, stable background, clean ending.
Use this promptReference workflow
Use image reference for [identity/style], video reference for [motion/camera rhythm], and audio reference for [beat/ambience]. Preserve [details]. One continuous shot.
Use this promptStart with the latest Wan model, then compare by usable output
Use the same shot goal across versions and compare consistency, retries, control, and credits—not a single attractive frame.
Create with Wan 3.0