WAN 3.0 Video

3 models available

Provider: Alibaba

Updated: Recently

WAN 3.0 is an All‑in‑One reference‑based video generation model that uniformly supports multiple use cases including text‑to‑video, image‑to‑video (first‑frame / first‑and‑last‑frame), and reference‑based video generation. It can generate videos of up to 30 seconds in length.

Available Models

WAN 3.0 Reference to Video

wan3.0-r2v

WAN 3.0 Reference‑based Video Generation supports inputting reference images, videos, audios, files or web links, and the model automatically understands intentions to generate videos.

WAN 3.0 Image to Video

wan3.0-i2v

WAN 3.0 Image to Video is an image‑to‑video variant of WAN 3.0. It supports inputting both the first and last frames to strictly define the starting and ending images of the video.

WAN 3.0 Text to Video

wan3.0-t2v

WAN 3.0 Text to Video is the text‑to‑video version model of WAN 3.0, which generates videos solely through prompts without importing any media files.

WAN 3.0 VIDEO / ALIBABA CLOUDOFFICIAL MODEL ADAPTER

A single model. Three creative inputs.

Make the next shot.

Wan 3.0 Video turns a prompt, a first frame, or a shelf of references into directed video. Bring the same production language from concept to API call.

Read the API docs

Need faster results? Try Wan 3.0 Prime, the high-speed tier

1080P

Output resolution

2–30s

Generation duration

5 ratios

Aspect ratios

Async

Execution mode

01 / Choose your source

Three model IDs. One production surface.

Pick the public ID that matches the assets in your pipeline. APIPod keeps provider-side naming and routing behind the same generation contract.

wan3.0-t2v

Text to video

Start with a prompt and direct the subject, camera, motion, lighting, and scene rhythm in natural language.

Prompt in · Video out

wan3.0-i2v

Image to video

Animate a first frame, or define a transition with first-and-last-frame control for a more deliberate shot.

First frame · Optional last frame

wan3.0-r2v

Reference to video

Combine reference images, videos, audio, one document file, or one public webpage to carry characters, objects, style, sound, and a complete brief into a new scene.

Images · Video · Audio · File/Link

02 / Directable by design

Control the shot without losing the spark.

Wan 3.0 is designed for real creative iteration: start broad, lock the visual anchors, then request the exact output your product needs.

One model across the brief

Move from a text-only idea to a reference-led sequence without changing providers or request conventions. Three APIPod IDs map to the appropriate Wan 3.0 workflow.

Reference media that stays in frame

Use up to 10 images, 5 videos, and 5 audio files in reference-to-video requests to anchor identity, composition, movement, and atmosphere.

Production-ready output controls

Choose 480P, 720P, or 1080P, set a 2-30 second duration, and use adaptive or five standard aspect ratios for the destination surface.

IMAGE 1

IMAGE 2

REF 3

REF 4

REFERENCE INPUTS / READY

03 / Bring the references

A visual brief can be more than one file.

Reference-to-video lets your prompt work alongside the source material. Build a consistent visual world from the pieces your team already has.

Images

Up to 10 reference images

Videos

Up to 5 reference videos

Audio

Up to 5 reference audio files

Files

1 document up to 100MB

Webpage

1 public link with no login

04 / Async by default

A clean path from brief to render.

Every request is an asynchronous task. Keep your product responsive while the provider renders, then retrieve the finished asset from the same task lifecycle.

Select a mode

Choose wan3.0-t2v, wan3.0-i2v, or wan3.0-r2v based on the inputs you have.

Compose the brief

Send your prompt with the frames, references, one optional document or public webpage, duration, resolution, and aspect ratio your shot needs.

Submit the task

APIPod returns a task ID while the Wan 3.0 adapter submits and monitors the asynchronous generation.

Retrieve the video

Poll the task until it completes, then pass the final video URL into your app, editor, or delivery pipeline.

Ship video with one stable API.

Use the same APIPod generation surface for all three Wan 3.0 modes. Your integration stays stable while provider routing stays internal.

Get API access

# Wan 3.0 text-to-video request
POST https://api.apipod.ai/v1/videos/generations
  Authorization: Bearer $API_KEY

{
  "model": "wan3.0-t2v",
  "prompt": "A cyclist cuts through a rain-lit city at blue hour, handheld camera, reflections on asphalt",
  "duration": 10,
  "resolution": "1080P",
  "aspect_ratio": "adaptive"
}

05 / Request contract

The details your pipeline can count on.

Open Alibaba Cloud reference

Resolution

480P / 720P / 1080P

1080P is the default output resolution.

Duration

2-30 seconds

Use -1 to let the model determine an intelligent duration.

Aspect ratio

Adaptive + 5 ratios

16:9, 4:3, 1:1, 3:4, and 9:16 are supported.

Reference media

10 images / 5 videos

Reference-to-video accepts up to 10 images and 5 videos.

Reference audio

Up to 5 files

Add up to 5 audio references to guide the generated scene.

Execution

Async task

Create a task, poll its status, and retrieve the completed video.

06 / Questions, answered

Before you press render.

What is Wan 3.0 Video?

Wan 3.0 Video is Alibaba Cloud's wan3.0-video model exposed through APIPod. It supports text-to-video, first/last-frame image-to-video, and reference-to-video workflows.

Which public model ID should I use?

Use wan3.0-t2v for text-only prompts, wan3.0-i2v when you have a first frame or first-and-last frames, and wan3.0-r2v when you want to guide the render with reference images, videos, audio, a document file, or a public webpage.

How many reference assets can I send?

The reference-to-video mode accepts up to 10 images, 5 videos, 5 audio files, plus either 1 document file or 1 public webpage per request; file and webpage inputs cannot be combined. The image-to-video mode uses a required first frame and can take an optional last frame.

What duration and resolution are supported?

The current contract supports 2-30 seconds, or -1 for intelligent duration selection. Resolution can be 480P, 720P, or 1080P, with 1080P as the default.

Can audio be used as a reference?

Yes. Wan 3.0 reference-to-video accepts audio references, one document file, or one public webpage. Webpages must use HTTP/HTTPS, be publicly reachable, and require no login. Supported document formats include PDF, DOCX, PPTX, XLSX, TXT, Markdown, and common Apple document formats; the file limit is 100MB.

Where can I read the upstream API reference?

See the official Alibaba Cloud Wan 3.0 Video API reference for the complete request schema, parameter constraints, and asynchronous task lifecycle: https://help.aliyun.com/zh/model-studio/wan3-video-generation-api-reference

READY WHEN THE BRIEF IS

Turn the next visual idea into a working endpoint.

Start with one prompt, add references when the shot needs them, and keep the rest of your production flow on a single API.

Same capabilities, faster turnaround — explore Wan 3.0 Prime