Alibaba Available on Video Blade

Wan 3.0 Video API

Text to videoImage to videoReference to video

Alibaba's Wan 3.0 Video makes longer, story-driven clips from a prompt or keyframes, with speech, music and ambience built in.

8-second clip from · per second

Live price in USD — you see the exact price before anything runs. Plans start at .

Use from your AI agent Get started

Input image 1 — Example output
Five keyframes become a sequence presenting Chinese-style headpieces, each character in a steady half-body portrait.
“Based on the uploaded 5 keyframe images, create a 10-second Chinese-style character headpiece showcase video. Maintain the refined, elegant, and restrained aesthetic of the reference video, with traditional Chinese visual elements and premium fashion presentation. The 5 characters appear sequentially in the upload order. Each character stays exactly centered in the frame with a stable half-body portrait composition. No position changes, no body movement, and no large camera motion. Each character remains visible for about 2 seconds. The main animation consists of subtle head movements: slight left or right turns, gentle head tilts, small chin lifts, and returning to the original pose. The movements should feel graceful and elegant, highlighting the details of the headpieces. Tassels, flowers, feathers, butterflies, magpies, kites, and other decorations can have delicate secondary motion following the head movement. The background should preserve Chinese-style patterns such as auspicious clouds, pavilions, towers, and landscape motifs, remaining mostly static with only subtle parallax or breathing motion. Between characters, use a consistent horizontal card flip transition. The current character image rotates horizontally like a vertical card around the center axis, briefly becoming narrow from the side before revealing the next character. Keep the character position fixed during transitions. Do not use particles, zoom effects, random movement, or characters entering from the edges. Camera style: locked-off premium portrait shot with only a very subtle push-in. The overall rhythm is: "Display hold → horizontal card flip → display hold → horizontal card flip" Final shot holds on the fifth character with a stable elegant pose.”

Input image 2 — Example output
Five keyframes become a sequence presenting Chinese-style headpieces, each character in a steady half-body portrait.
“Based on the uploaded 5 keyframe images, create a 10-second Chinese-style character headpiece showcase video. Maintain the refined, elegant, and restrained aesthetic of the reference video, with traditional Chinese visual elements and premium fashion presentation. The 5 characters appear sequentially in the upload order. Each character stays exactly centered in the frame with a stable half-body portrait composition. No position changes, no body movement, and no large camera motion. Each character remains visible for about 2 seconds. The main animation consists of subtle head movements: slight left or right turns, gentle head tilts, small chin lifts, and returning to the original pose. The movements should feel graceful and elegant, highlighting the details of the headpieces. Tassels, flowers, feathers, butterflies, magpies, kites, and other decorations can have delicate secondary motion following the head movement. The background should preserve Chinese-style patterns such as auspicious clouds, pavilions, towers, and landscape motifs, remaining mostly static with only subtle parallax or breathing motion. Between characters, use a consistent horizontal card flip transition. The current character image rotates horizontally like a vertical card around the center axis, briefly becoming narrow from the side before revealing the next character. Keep the character position fixed during transitions. Do not use particles, zoom effects, random movement, or characters entering from the edges. Camera style: locked-off premium portrait shot with only a very subtle push-in. The overall rhythm is: "Display hold → horizontal card flip → display hold → horizontal card flip" Final shot holds on the fifth character with a stable elegant pose.”

What Wan 3.0 Video is good for

Wan 3.0 Video is Alibaba's all-in-one model for longer, story-driven clips. It accepts up to 10 reference images alongside text, video, audio, and keyframes, and keeps characters, clothing, props, and settings close to their references as scenes change; it can also edit or extend existing footage. Generated speech, music, and ambient sound follow the action, which suits branded stories, fashion pieces, and narrative shorts.

Longer story clips with room for setup, action, and a payoff
Fashion and product clips that start from your own photo
Sequences anchored by opening and closing keyframes
Story clips with speech, music and ambient sound generated together

Wan 3.0 Video pricing

Option	Price, USD	Video Blade tokens
480p		tokens
720p		tokens
1080p		tokens

Prices are per second and live from our price list. Tokens are held when a run starts and settled to what was actually used; a failed run costs nothing.

Use Wan 3.0 Video from your AI agent

Create an account on Video Blade and pick a plan — from .
Copy the agent link from your dashboard into Claude Code, Codex or Cursor.
Say the phrase below — the agent shows the price, runs Wan 3.0 Video and saves the result.

Tell your agent

Create an 8-second 16:9 1080p clip with Wan 3.0 Video from my lobby photo as the first frame: guests check in at a sunlit hotel, the camera glides toward the pool, soft lounge music.

This request ≈ · 1080p · 8 s

Wan 3.0 Video FAQ

How much does Wan 3.0 Video cost?

8-second clip from · per second. Prices are live from our price list, and you see the exact price before anything runs.

What kinds of input can Wan 3.0 Video work from?

Besides a text prompt, Wan 3.0 Video can take up to 10 reference images, video clips that show motion or camera style, audio for voice and rhythm, and keyframes that set how a sequence opens or ends. It can also draw on briefs, presentations, PDFs, and webpages as source material.

How closely does Wan 3.0 Video stick to its references?

Holding references steady is a core focus. Characters, clothing, props, locations, audio cues, and even the spatial relationships between objects are kept close to the supplied material as actions and scenes change, which matters for recurring characters, product details, and brand assets that cannot drift.

How do I use Wan 3.0 Video from my AI agent?

Create a Video Blade account, paste the agent link from your dashboard into Claude Code, Codex or Cursor, and ask: “Create an 8-second 16:9 1080p clip with Wan 3.0 Video from my lobby photo as the first frame: guests check in at a sunlit hotel, the camera glides toward the pool, soft lounge music.” The agent runs Wan 3.0 Video (model id wan-3-0-video), shows the price before it starts and saves the result to your dashboard.

How is Wan 3.0 Video billed?

By the second of video: the rate depends on the option you pick (mode, resolution, sound), and a clip costs the rate times its length. Tokens are held when a run starts and settled to what was actually used; a failed run costs nothing. Current prices are shown on this page.

What settings does Wan 3.0 Video support?

Resolution: 480p, 720p or 1080p. Aspect ratio: adaptive, 16:9, 4:3, 1:1, 3:4 or 9:16.

Wan 3.0 Video in the Video Blade API

wan-3-0-video

Your agent sends this model id with the order; the result lands in the project, and every request is in your generation log with its price.

Through Video Blade a clip is ordered from a prompt with an optional start frame (and an end frame where the model takes one), in the lengths and resolutions the model offers.

POST /api/v1/projects/{id}/video-sessions/{item_id}/generate
{"prompt": "…", "model": "wan-3-0-video", "duration": …, "start_url": "https://…"}

Parameters and API reference →Getting started →

Similar models

ByteDance

Wan 3.0 Video API

What Wan 3.0 Video is good for

Wan 3.0 Video pricing

Use Wan 3.0 Video from your AI agent

Wan 3.0 Video FAQ

Wan 3.0 Video in the Video Blade API

Similar models

Seedance 2.5

Wan 3.0 Video Prime

Seedance 2.0

Veo 3.1