Longer single takes
Generate up to 30 seconds in one pass. Complete a narrative beat, product demo or social clip without stitching fragments together.
Alibaba Tongyi Wan AI Video
Tell longer stories with Wan 3.0
Turn prompts, first and last frames, and up to 20 reference assets into expressive AI video with native audio. Generate 2 to 30 second clips in one pass with adaptive framing and cinematic camera control.
Text to Video / Image to Video
Wan 3.0first_frame_url
Click to upload or drag and drop
last_frame_url (Optional)
Click to upload or drag and drop
Describe the shot. Type @ to reference attached media (e.g. @Image1, @Video1, @Audio1). Optional when media is attached.
Click or drop images here
Up to 10 images (JPG/PNG/WEBP), max 20MB eachClick or drop videos here
Up to 5 videos (MP4/MOV), max 100MB each, 15 seconds total; reference video time + output duration must stay within 30sClick or drop audio here
Up to 5 audio files (MP3/WAV), max 15MB each, 15 seconds total. Audio requires at least one reference image or video.Estimated cost 150 gems
Your video will appear here
What is Wan 3.0?
Turn prompts, first and last frames, and up to 20 reference assets into expressive AI video with native audio. Generate 2 to 30 second clips in one pass with adaptive framing and cinematic camera control.
2–30sSingle-pass duration
480p / 720p / 1080pThree quality tiers
6 ratiosAdaptive and social-ready
20 assetsReference images, videos, audio
Why creators use Wan 3.0
From a first idea to a finished clip, the workflow stays expressive and practical.
Rich reference understanding
Feed up to 10 reference images, 5 reference videos and 5 audio files. Characters, styles, camera moves and sound direction stay aligned with your assets.
Frame-accurate control
Pin the opening and closing frames of your shot. Wan 3.0 handles the motion in between while respecting your composition.
Native audio in one pass
Dialogue, ambience and effects are generated with the picture, so lip movement and sound land in sync — no separate sound pass.
Cinematic camera language
Describe push-ins, orbits and tracking shots in natural language. Motion stays stable and perspective holds through the move.
Every format you publish
Adaptive framing follows your source media, or lock 16:9, 9:16, 1:1, 4:3 or 3:4 for the channel you target.
How to create a video
Steps from idea to downloadable result.
Write the direction
Describe the subject, action, camera and sound. Reference your attached assets with @Image1, @Video1 or @Audio1 tokens in the prompt.
Add visual or audio guidance
Upload a first frame with an optional last frame, or attach reference images, videos and audio. Choose one workflow — frames and references cannot be combined.
Set the shot
Pick resolution, aspect ratio and duration from 2 to 30 seconds. The studio estimates your cost live as you configure the shot.
Generate and review
Start the generation and watch the preview panel. Your clip is saved to My Creation, so you can close the page while it renders.
Iterate with intent
Refine the prompt, adjust duration or swap references. Longer takes benefit from beat-by-beat direction in the prompt.
Publish your best take
Download the finished MP4 and ship it — vertical for shorts, widescreen for the web, square for feeds.
Wan 3.0 FAQ
Answers to common questions about inputs, settings and billing.
What is Wan 3.0?
Wan 3.0 is Alibaba’s next-generation all-in-one video model. It generates video with native audio from text prompts, first and last frame images, and multimodal references including images, video clips and audio files.
What inputs does Wan 3.0 support?
You can start from a text prompt alone, add a first frame (optionally with a last frame), or use reference media: up to 10 reference images, 5 reference videos and 5 reference audio files in a single request.
How long can Wan 3.0 videos be?
Clips run from 2 to 30 seconds in one generation. When reference videos are attached, the total reference video duration plus your output duration must stay within 30 seconds.
Does Wan 3.0 generate audio?
Yes. Wan 3.0 produces a native audio track — dialogue, ambience and sound effects — in the same pass as the picture. You can switch audio off if you only need silent footage.
Can I combine first/last frames with reference media?
No. The frames workflow and the reference-media workflow are mutually exclusive, following the provider API rules. Reference audio also requires at least one reference image or video.
How does billing work?
Cost is a flat per-second rate set by your chosen resolution (480p, 720p or 1080p) multiplied by the output duration. Audio does not change the price, and reference media does not add to the billed seconds.
Ready to direct your next shot?
Start with a prompt or upload a frame and create a 30-second story with Wan 3.0 today.
Start creating with Wan 3.0