Wan 3.0 is Alibaba's video generation model for native 30-second storytelling, up to 1080P output, native audio-visual generation, omni reference input, and high-fidelity real-world continuity.

How long can Wan 3.0 videos be?

A single generation can run for up to 30 seconds and support output up to 1080P, giving longer scenes room for complete narrative structure and camera progression.

What files can Wan 3.0 use?

Wan 3.0 supports images, text, video, audio, documents, and web pages. The official release describes combinations of up to 10 images, 5 videos, and 5 audio clips as reference input.

What can Wan 3.0 keep consistent?

It is designed to preserve identity, actions, props, sound, space, visual style, software UI, charts, and visible text so each frame remains credible and connected to the brief.

Can I generate videos for free with Wan 3.0?

During the current ArtArch promotion from August 24 to September 7, each user can generate 100 videos for free. Duration and parameters are not limited during the promotion.

Can Wan 3.0 connect to production workflows?

Yes. The Wan 3.0 Video API is officially available for developers to move from creative testing into advertising, film, games, design, and content production workflows.

Wan 3.0

Create AI videos online with Wan 3.0. Generate a native 30-second sequence up to 1080P with synchronized sound, and direct it with images, text, video, audio, documents, or web references. During the current promotion, each user can generate 100 videos for free.

Create with Wan 3.0 CLI & Skills

Single generationNative 30 sec

OutputUp to 1080P

Audio / videoNative generation

References10 images + 5 videos + 5 audio

Turn your existing materials into complete videos.

Start with a product image, a written brief, or a character reference. Wan 3.0 turns the details you provide into connected shots with clear action and continuity.

Generated technology scene used as a product story concept frame

Product image → product video

Generate a product ad from one image

Upload a product image and keep its shape, materials, logo, and user interaction consistent across the video.

Generated landscape concept frame for a story directed by a written brief

Written brief → story video

Turn a written brief into a story

Describe the goal, setting, characters, and key beats. Wan 3.0 arranges them into a video with a clear beginning, middle, and ending.

Generated atmospheric scene representing a fictional narrative concept

Character reference → connected shots

Keep characters and settings consistent

Start with a character reference and preserve the face, clothing, props, and space across a continuous story or series.

Everything the scene needs, in one creative brief.

Wan 3.0 upgrades duration, reference breadth, audio-visual generation, and real-world fidelity in one workflow.

Native 30-second generation

Build a complete narrative in one output, with longer camera movement, one-take sequences, and multi-beat pacing that can unfold naturally.

Omni reference and creation

Combine up to 10 images, 5 videos, and 5 audio clips, then add text, documents, or web pages to carry the details behind the brief.

Native audio-visual generation

Generate sound, dialogue, rhythm, and lip-sync as part of the scene, so the picture and soundtrack move together from the first output.

Real-world fidelity

Keep identity, actions, props, spaces, visual style, software UI, charts, and visible text accurate and consistent across the sequence.

Built for production

Move from idea to finished direction for film, advertising, design, games, and cultural content, with Wan 3.0 Video API access now open.

Creative document workflow representing a Wan 3.0 production brief

ArtArch workflow

Direct a complete 30-second sequence.

Give each reference a clear job, then let Wan 3.0 carry the story, sound, and continuity through the final frame.

InputAdd images, text, video, audio, documents, or web references.

DIRECTDefine the characters, space, camera movement, dialogue, and final beat.

GENERATERender the full 30-second sequence with native sound and up to 1080P output.

REVIEWCheck identity, actions, space, visible text, and audio-visual sync before delivery.

Give every reference a clear responsibility.

Wan 3.0 reads visual, verbal, sonic, structured, and web context together, so the source material can carry the details the model should preserve.

Source	Use it to control	Best production use
Images	Identity, product shape, composition, style.	Characters, products, environments, and continuity.
Text	Action, camera language, pacing, dialogue.	Original direction and fast creative exploration.
Video + audio	Motion, rhythm, performance, sound intention.	Reference-led shots, music visuals, and lip-sync.
Documents + web	Requirements, structured facts, UI, and visible text.	Brief-led production, knowledge content, and product demos.

Separate what changes from what stays.

Wan 3.0 can follow a large multimodal brief. Separate identity, action, sound, and output requirements so each input has a clear job.

Wan 3.0 prompt example

Source — Use @Image1 for the male character, @Image2 for the female character, and @Image3 for the dojo space.

STORY — A 30-second confrontation that moves from a quiet two-shot into fast hand-to-hand action, then settles on a shared final look.

AUDIO — Generate natural dialogue, room tone, impact sounds, and synchronized lip movement as part of the scene.

KEEP — Preserve both identities, clothing, dojo layout, props, camera axis, and visual style across every beat.

Output — Use 1080P with an adaptive ratio. Review faces, actions, spatial relationships, dialogue, and visible text before export.

Wan 3.0 FAQ

Answers based on the public Wan 3.0 release information and the current ArtArch promotion.

Start with free video generation. Scale to production.

Try Wan 3.0 during the current promotion, then take the same reference-led workflow into ArtArch Studio.

Start creating