Wan 3.0 — 30-Second AI Video from Text, Image & Reference30 Seconds
Alibaba Tongyi Lab's Wanxiang 3.0 video model. Generate up to 30 seconds in a single pass at 480p, 720p, or 1080p — from a prompt, a first frame (optional last frame), or mixed image, video, and audio references — with optional native audio on SeeVid.
Available on SeeVid platform
Alibaba Tongyi's Wanxiang 3.0 Video Model
Wan 3.0 (Tongyi Wanxiang 3.0) is the 2026 generation of Alibaba Tongyi Lab's AI video family. Public materials describe up to 30 seconds in one generation, native audio-visual output, and an omni-reference system for text, images, video, and audio. On SeeVid you can run text-to-video, image-to-video (first frame, optional last frame), and reference-to-video with up to 10 images, 5 clips, and 5 audio files.
Up to 30 Seconds, One Pass
Choose 2–30 seconds or Auto. Auto lets the model pick a length that fits the prompt instead of padding a short beat or cutting a longer story. No stitching of many 5-second clips for a single take.
Text, Image & Mixed References
Start from a prompt, animate a still (optional end frame), or mix up to 10 images, 5 videos (1–15s each, 15s total), and 5 audio files.
Native Audio in the Same Pass
Speech, singing, effects, and ambience can be generated with the picture. Audio is on by default; turn it off for a silent clip. Lip-sync and multilingual dialogue are headline capabilities of the Wan 3.0 family.
Faces, Props, Space & Style
Reference-to-video is built for character, product, space, and style consistency across a longer clip — useful for ads, explainers, and short narrative drafts before you polish in an editor.
Longer Takes, Stronger References, Audio Included
Wan 3.0 targets creators who need more than a 5–15 second beat: a 30-second narrative canvas, fine-grained reference locking, and sound that ships with the picture.
Wan 3.0 on SeeVid
Exact options match the AI video generator. Credits are billed per output second (Auto counts as 20s). Reference-to-video also bills reference-video seconds.
Text-to-Video
Describe scene, camera, style, and dialogue. Duration 2–30s or Auto; 480p / 720p / 1080p; aspect adaptive, 16:9, 9:16, 1:1, 4:3, or 3:4.
Image-to-Video
One image is the first frame; add a second image as the last frame. Max two images. Same duration, quality, and aspect options as text-to-video.
Reference-to-Video
Up to 10 images (≤20MB each), 5 videos (1–15s, ≤100MB, 15s total), and 5 audio files (1–15s, ≤15MB, 15s total). Output plus reference video cannot exceed 30 seconds.
Smart Duration
Pick an exact length or Auto. Auto is billed as a 20-second job so a short product beat is not padded and a longer story is not cut mid-beat.
480p / 720p / 1080p
Three quality tiers. SeeVid credits: 3.7 / 7.5 / 15 per output second at 480p / 720p / 1080p. Example: 720p × 5s = 37 credits.
Native Audio Toggle
Audio is on by default (speech, effects, ambience). Turn generate audio off for a silent clip you will score later.
Frequently Asked Questions
Common questions about Wan 3.0 (Tongyi Wanxiang 3.0) and using it on SeeVid.
Start Creating with Wan 3.0 Today
Try Wan 3.0 on SeeVid — text-to-video, image-to-video and reference-to-video, 2–30 second clips with optional native audio at up to 1080p.