Wan 3.0 AI Video Generator

Create a Wan 3.0 video from a prompt, first and last frames, or reference images, video, and audio. Choose 480P, 720P, or 1080P and see the credit cost before you generate.

Generate with Wan 3.0

Prompt

0/2500

Public Visibility

Your Wan 3.0 Videos

Wan 3.0 is live. Choose Text, Frame, or Reference mode; 720P is the default, and the exact Nana Banana credit cost appears before generation.

Official Wan 3.0 examples

Watch real Wan 3.0 outputs published by Alibaba

Alibaba published these videos as Wan 3.0 examples. Nana Banana serves web-optimized copies from its CDN; they are official examples, not results generated here.

Open the official Wan 3.0 showcase

Official sample · 720P · 30s

Sparring in the ruins

Two image references keep a tabby cat and a wolf consistent through fast action, changing camera angles, fur, clothing, and spatial movement.

Official sample · 720P · 24s

Reality and simulation meet in a dojo

Three reference images guide two characters and a dojo, including character movement, Japanese dialogue, and synchronized sound.

Official sample · 720P · 25s

A brand document becomes a warm café ad

A brand document provides the story, ingredients, and setting. One creative direction turns that material into a complete café commercial.

Wan 3.0 capabilities

Create longer videos from more kinds of input

Wan 3.0 combines prompts, media references, documents, and native audio in one video workflow.

Tell a story for up to 30 seconds

Choose any whole-second length from 2 to 30, or let smart duration pick a length from your prompt. A longer shot gives actions, camera moves, and dialogue room to finish.

Use more than a text prompt

Wan 3.0 can read text, images, video, audio, office files, and a public web page. A slide deck or report can become the starting point for a video.

Create picture and sound together

Audio is on by default. The model can build voices, music, and scene sound together with the visuals, which helps speech and action feel connected.

Keep references recognizable

You can guide a result with up to 10 images, 5 short videos, and 5 short audio clips. These references help hold onto a person, product, place, style, or voice.

Control the start and finish

Use a first frame to decide how a clip begins, or add both a first and last frame to guide where the motion should end.

Choose the right Wan 3.0 route

Use Text mode for a prompt, Frame mode for a strict first frame and optional last frame, or Reference mode for images, video, and audio. Matching the route to the input prevents rejected requests.

Official specifications

Plan with Wan 3.0's API limits

Use these published limits to choose the right duration, reference media, aspect ratio, and output quality.

Model ID: wan3.0-text-to-video, wan3.0-image-to-video, and wan3.0-reference-video
Video length: 2–30 seconds without a video input; smart duration is available. Input video time plus output time cannot exceed 30 seconds.
Output quality: 480P, 720P, or 1080P.
Aspect ratios: Adaptive, 16:9, 4:3, 1:1, 3:4, and 9:16.
Reference media: Up to 10 images, 5 videos, and 5 audio clips. Reference videos and audio each have a 15-second total limit.
Files and web pages: One file or one public link per request. Files can be up to 100 MB and 50 pages, with common document, slide, spreadsheet, PDF, text, and Markdown formats supported.
Audio: Included by default and can be turned off. The price does not change when audio is disabled.
API workflow: Nana Banana sends the request through EvoLink's unified video API, polls the task until it finishes, and saves the completed video to persistent storage.

Prompt workflow

Build a better Wan 3.0 prompt

Describe the scene in order, add only the references that matter, and choose settings for the final destination.

1
Describe one clear scene
Say who or what appears, what happens, how the camera moves, what the light looks like, and what should be heard. Put events in time order.
2
Add only helpful references
Use an image for appearance, a short video for motion, or audio for a voice. Name each reference in the prompt so its job is easy to understand.
3
Pick length, shape, and quality
Choose a duration that fits the action, a horizontal or vertical ratio for the destination, and a lower resolution for drafts before a 1080P final.

Model access

Wan 3.0 availability

Wan 3.0 is available now in Nana Banana's video generator through EvoLink's production API.

Create with Wan 3.0 here

The generator above uses Wan 3.0 directly for text-to-video, first-and-last-frame video, and multimodal reference video.

Clear credit pricing

Wan 3.0 is priced at 1.5 times the current EvoLink upstream cost. The generator converts that amount to whole Nana Banana credits and shows it before you submit.

Output is saved for you

EvoLink result links expire, so Nana Banana copies a successful result to its own storage before showing it in your creations.

Wan 3.0 facts last checked: August 26, 2026

Official resources

Wan 3.0 documentation

Review the official model resources and EvoLink API information used for this page and integration.

Wan 3.0 official model releaseDemos, 30-second output, supported inputs, availability, and pricing.Wan 3.0 Model Studio guideA plain overview of longer videos, document input, editing, use cases, and known limits.Wan 3.0 API referenceExact model ID, media limits, duration, resolution, ratios, audio, and task status rules.EvoLink Wan 3.0 API and pricingModel IDs, request parameters, output limits, and current upstream pricing used by Nana Banana.

FAQ

Wan 3.0 FAQ

Answers about the model, access, pricing, inputs, and output.

What is Wan 3.0?

Wan 3.0 is Alibaba's newest all-in-one AI video model. It makes videos from text or references and can generate up to 30 seconds of visuals and sound in one task.

Is Wan 3.0 available on Nana Banana?

Yes. Choose Wan 3.0 in the generator above, then use Text, Frame, or Reference mode. Nana Banana handles the EvoLink task and stores the completed result.

Is Wan 3.0 free?

New accounts can try it with available signup credits. Each generation uses credits based on duration and resolution; Nana Banana sets the price at 1.5 times the current EvoLink upstream cost and shows the total before generation.

How long can a Wan 3.0 video be?

Without a video input, choose any whole-second duration from 2 to 30. Smart duration can choose for you. When a video is used as input, input time plus output time must stay within 30 seconds.

Does Wan 3.0 generate audio?

Yes. Audio is enabled by default, so Wan 3.0 can create dialogue, voices, music, and scene sound with the picture. You can also turn audio off.

What can I upload to Wan 3.0?

The API accepts text, images, short videos, short audio clips, a document, or a public web page. It also supports first-frame and first-plus-last-frame image control.

What resolution does Wan 3.0 support?

The official API lists 480P, 720P, and 1080P output, with horizontal, square, vertical, and adaptive aspect ratios.

Is Wan 3.0 open source?

Alibaba's official sources describe Wan 3.0 as a cloud model and API, not a downloadable open-weight release. Older Wan versions have open repositories, but that does not make Wan 3.0 open source.

Create your Wan 3.0 video

Write a prompt, choose a mode and output settings, then generate directly with Wan 3.0.