Commercial
Wan 3.0 PrimeWan 3.01080p. Create, edit, extend.
Generate, edit, and extend complete AI video with native audio, precise frame control, and up to twenty multimodal references.
2–30soutput duration
1080pmaximum resolution
20combined references
Nativeaudio generation
Interactive Wan 3.0 workspace
Direct the model, not just the prompt.
Explore the actual Wan 3.0 controls, assemble a shot, then open the complete generation workspace in Unify.
Generate
Create a new sequence from text, frames, or a multimodal reference pack.
Source setup
Prompt-only generation
No source media required. Describe subject, motion, camera, light, and sound below.
206 / 20,000
Model intelligence
Wan 3.0 Prime1080p16:9 · 10sAudio on
Continue in the Wan workspace This page previews the workflow. Source files are selected securely again inside the generator.
20 combined references First + last frame Native sound 30 fps output
Creative direction board
One model. Every kind of motion.
Explore the visual formats available across Unify, then use Wan 3.0’s text, frame, video, image, and audio controls to direct your own version.
Consistent motion
Controlled continuity
Complete capability map
Create. Control. Transform.
Wan 3.0 uses one model for new generation, multimodal direction, editing, and extension—without splitting the workflow across separate tools.
Generate
Text to video
Create a complete scene from a prompt, including subject, action, camera, pacing, and sound.
Reference
Image-guided video
Use reference images to preserve a person, product, environment, or visual language.
Frames
First and last frame
Define the opening frame alone or pair it with an ending frame for controlled transitions.
Multimodal
Up to 20 references
Combine as many as 10 images, 5 short videos, and 5 audio clips in one reference-led brief.
Sound
Native audio
Generate dialogue, background music, ambience, and sound effects with the moving image.
Edit
Prompt-led video editing
Upload a source video and describe the transformation while retaining the material you want to keep.
Extend
Video extension
Continue an existing video with adaptive framing and model-selected duration based on the source.
Context
File input
Supply one supported file as source context when a prompt alone is not enough.
Context
Web link input
Use one supported link as source context instead of uploading a file.
Output
480p to 1080p
Move from efficient 480p drafts to 720p or full-HD 1080p delivery.
Format
Six aspect modes
Choose adaptive, 16:9, 4:3, 1:1, 3:4, or 9:16 for film, feeds, and vertical video.
Duration
2 to 30 seconds
Select any whole-second duration across the supported range, rendered at 30 frames per second.
Inputs and output
A larger creative brief.
Mix the sources a production actually uses while keeping every hard limit visible before upload.
Reference images
Up to 10
JPEG, PNG, WebP, or BMP · 20 MB each
Reference videos
Up to 5
MP4 or MOV · 15 seconds and 100 MB each
Reference audio
Up to 5
MP3 or WAV · 15 seconds combined · 15 MB each
Combined references
Up to 20
Images, video, audio, and one file or link
Prompt
Up to 20,000 characters
Prompt extension is enabled by default
Output
2–30 seconds
480p, 720p, or 1080p · 30 fps
Edit and extend
The first render is only the start.
Upload a selected take, isolate the change in words, or continue the next beat from the same source video. Wan 3.0 keeps generation, editing, and extension inside one model.
Four-step workflow
From brief to continuation.
Generate a new scene
Start from text, a first frame, a first-and-last-frame pair, or a set of references. Name each source in the prompt and describe exactly what it controls.
Direct motion and sound
Describe the action, camera path, timing, dialogue, ambience, music, and sound effects. Wan 3.0 can resolve picture and native audio together.
Edit the selected take
Switch to Edit, provide the source video, and isolate the requested change. State what must remain untouched to protect continuity.
Extend the sequence
Switch to Extend and describe the next beat. Wan follows the source framing and selects a compatible continuation length automatically.
Wan 3.0 prompts
Write for motion, continuity, and sound.
01
First-and-last-frame product reveal
Use Image 1 as the first frame and Image 2 as the final frame. Begin on an extreme macro of condensation moving across the bottle, then pull back in one smooth arc as the studio light changes from cool blue to warm amber. End precisely on Image 2. Premium commercial lighting, realistic reflections, soft room tone and a restrained bass hit at the reveal.
02
Multimodal character scene
Keep the lead character identical to Reference Image 1 and use Reference Video 1 only for walking rhythm. She crosses a windswept train platform at night, coat moving naturally, then stops as the train arrives behind her. Slow handheld push-in, realistic weight, distant rail ambience, no dialogue, preserve face and wardrobe in every frame.
03
Edit an existing advertisement
Edit the supplied video. Keep the camera movement, product position, hands, timing, and soundtrack unchanged. Replace only the environment with a minimalist gallery made from pale concrete and brushed steel. Match the original shadows and reflections so the product remains grounded in the shot.
FAQ
Everything before render.
What is Wan 3.0?
+
Wan 3.0 is Alibaba's all-in-one multimodal AI video model for text-to-video, frame-guided generation, reference-led generation, video editing, extension, and native audio. Unify provides the Wan 3.0 Prime quality tier in its video workspace.
How long can Wan 3.0 videos be?
+
A generated output can be any whole-second duration from 2 to 30 seconds. When reference video is supplied, its duration and the requested output must fit within the model's 30-second combined budget.
Which resolutions and aspect ratios does Wan 3.0 support?
+
Wan 3.0 supports 480p, 720p, and 1080p. Available aspect modes are adaptive, 16:9, 4:3, 1:1, 3:4, and 9:16.
Can Wan 3.0 generate audio?
+
Yes. Audio is enabled by default and can include dialogue, background music, ambience, and sound effects generated with the video.
How many references can I use?
+
A request can contain up to 20 references overall: as many as 10 images, 5 short videos, and 5 audio clips, plus the supported file-or-link input rules. First/last frame mode cannot be mixed with the multimodal reference mode in the same request.
Can Wan 3.0 edit or extend a video?
+
Yes. Edit and Extend are explicit modes in Unify. Both require a source video and use adaptive framing with smart duration so the model can follow the supplied footage.
Can I try Wan 3.0 on Unify?
+
Yes. Open the Unify video generator from this page and Wan 3.0 Prime will already be selected. The current credit quote is shown before submission.
Direct the whole sequence.
Choose Wan 3.0 Prime in Unify and move from prompt to reference-led generation, editing, and extension.