Wan 3.0 AI Video Generator and API

Specification	Credits	Price (USD)
480P	4 per second	$0.04 per second
720P	8 per second	$0.08 per second
1080P	16 per second	$0.16 per second

What Wan 3.0 (Wanxiang) generates

Wan 3.0, also searched as Wanxiang 3.0, is Alibaba's all-in-one video model on AIReiter. One model ID covers text, first/last frames, mixed image-video-audio references, and document or webpage to video, at 480p, 720p, or 1080p for 2 to 30 seconds.

Three generation modes

Text/Reference starts from a prompt and can add images, videos, and audio. First/Last Frame animates a required first frame, with an optional last frame. File/Webpage turns a document or public webpage into a clip.

Up to 30 seconds

Pick any whole-second length from 2 to 30. If you attach a reference video, input seconds plus output seconds cannot exceed 30.

Native audio by default

Clips include a soundtrack unless you turn audio off. Turning audio off does not change the per-second price.

Same controls in the playground and API

The fields on this page map to wan3_0 parameters on POST /api/openapi/submit.

AIReiter-tested Wan 3.0 results

These clips were generated through the AIReiter OpenAPI on August 24, 2026. Durations, pixels, and credits are from the delivered files, not a spec sheet.

Text to video at 720p, 5 seconds

Prompt: A red paper airplane glides slowly across a bright white studio, fixed camera, soft daylight, cinematic, no text.
Request: Text/Reference, 720p, 16:9, 5 seconds, audio on.
Delivered file: H.264 MP4, 1280x720, 5.04 seconds, AAC stereo.
Billing at the current 720p rate: 40 credits (8 × 5). Task time was about 4 minutes 12 seconds.

First/last frame from one still

First frame: a 1024x1024 still. Last frame was omitted.
Prompt: The scene comes to life with a slow push-in, cinematic lighting, no text.
Request: First/Last Frame, 480p, auto ratio, 2 seconds, audio on.
Delivered file: H.264 MP4, 632x632, 2.02 seconds, AAC stereo.
Billing at the current 480p rate: 8 credits (4 × 2).

Reference video billed with output seconds

Reference: the 2.02-second 480p paper-airplane clip, passed as Video 1.
Request: Text/Reference, 480p, 16:9, 2 seconds of new output, audio on.
Delivered file: H.264 MP4, 832x480, 2.02 seconds, AAC stereo.
Billing at the current 480p rate: 20 credits. The 2.02-second reference is ceiled to 3 seconds, then 4 × (2 + 3). Failed tasks are not charged.

What to make with Wan 3.0

Choose the mode that matches the material you already have, not a longer prompt.

Storyboard a shot

Upload a first frame, optionally a last frame, and describe the motion between them. Use this when the opening picture is already designed.

Turn a brief into a clip

File/Webpage accepts PDF, PPT, Word, text, markdown, or a public webpage. The prompt is optional; a short instruction still helps.

Keep a product or voice consistent

In Text/Reference, add up to 10 images, 5 videos, and 5 audio clips. Name them in the prompt as Image 1, Video 1, or Audio 1.

Cut for social formats

Lock 9:16, 1:1, or 16:9 instead of auto when the publishing channel is already decided.

Which Wan 3.0 mode to use

The three modes cannot be mixed in one request. First/last frames are a separate family from reference images, videos, audio, documents, and webpages.

Text / Reference

Use a prompt alone, or add images, videos, and audio. Best default when you are starting from language or mixed media.

First / Last Frame

Required first frame, optional last frame. Output framing follows the still. Do not send reference videos or documents in this mode.

File / Webpage

One document or one public webpage URL. Supported files include DOC(X), XLS(X), PPT(X), PDF, TXT, KEY, Pages, Numbers, and MD, up to 100MB and 50 pages. Documents and webpages are not billed.

Wan 3.0 parameters

The playground and API expose the same validated controls. Live prices are in the table on this page.

Resolution

480p, 720p, or 1080p. 720p is the default. Higher resolution costs more per second.

Duration

Any integer from 2 to 30 seconds. With a reference video, input plus output cannot exceed 30 seconds.

Aspect ratio

auto, 16:9, 9:16, 4:3, 3:4, or 1:1. Auto follows the reference or prompt.

Reference media

Up to 10 images (shortest side ≥ 240px, longest side ≤ 8000px, 20MB). Up to 5 videos (1–15s each, 15s total, MP4/MOV, 100MB). Up to 5 audio clips (1–15s each, 15s total, WAV/MP3, 15MB). Only reference video seconds are added to the bill.

How Wan 3.0 is billed

Charge = (output seconds + ceiled reference-video seconds) × the live per-second rate in the table on this page. Images, audio, documents, and webpages are free. Audio on or off does not change the rate. Failed tasks are not charged.

Worked example, 10 seconds at 1080p

The 1080p rate is 16 credits per second ($0.16/s), so 10 seconds cost 160 credits with no reference video.

Worked example, reference video

A 2-second 480p request with a 2.02-second reference clip is 20 credits: the reference is ceiled to 3 seconds, then 4 × (2 + 3).

Default 720p, 5 seconds

A 5-second 720p clip is 40 credits (8 × 5).

How to generate with Wan 3.0

Use the same steps in the playground or through the API.

Pick one mode

Text/Reference, First/Last Frame, or File/Webpage. Do not mix first/last frames with reference media, documents, or webpages.

Describe the motion first

Say who or what moves, then camera, lighting, and style. Name uploads as Image 1, Video 1, or Audio 1.

Set length and resolution

Choose 2–30 seconds and 480p, 720p, or 1080p. Higher resolution costs more; mute does not.

Submit and poll

POST /api/openapi/submit returns estimated credits immediately. Poll POST /api/openapi/query with the same out_task_id until completed or failed.

Call Wan 3.0 through the AIReiter API

All three modes share one asynchronous contract.

Model ID

Send wan3_0. Select the mode with params.type: all_reference, first_last_frame, or file_web.

Submit

POST /api/openapi/submit with model, params, and a unique out_task_id.

Query

POST /api/openapi/query with the same out_task_id. Completed responses include credits_used and the output video URL.

File and webpage fields

For File/Webpage send file_url or link_url, not both, and not with first/last frames or reference media.

Wan 3.0 FAQ

Common questions about Wanxiang 3.0 length, price, modes, documents, and the API.

/ 01

What is Wan 3.0 (Wanxiang 3.0)?

Wan 3.0 is Alibaba's all-in-one video model. On AIReiter it generates 2–30 second clips at 480p, 720p, or 1080p from text, first/last frames, mixed references, documents, or webpages.

/ 02

How is Wan 3.0 different from earlier Wan video models?

This page exposes three modes on one model ID, native audio by default, up to 30 seconds, and document or webpage input. Earlier Wan pages on AIReiter do not combine those controls.

/ 03

How long can Wan 3.0 generate?

Any integer from 2 to 30 seconds. With a reference video, input seconds plus output seconds cannot exceed 30.

/ 04

How much does 10 seconds at 1080p cost?

Use the live 1080p per-second rate in the table on this page. The current rate is 16 credits/second ($0.16/s), so 10 seconds cost 160 credits with no reference video.

/ 05

Does turning audio off make Wan 3.0 cheaper?

No. Audio is on by default. Mute does not change the per-second price.

/ 06

Can I mix first/last frames with a reference video?

No. First/last frames cannot be combined with reference images, videos, audio, documents, or webpages in the same request.

/ 07

Can I turn a PPT or webpage into a video?

Yes. Use File/Webpage with a document URL or a public webpage. Prompt is optional. Those inputs are not billed; you still pay for output seconds.

/ 08

Which API model ID should I use?

Use wan3_0 with POST /api/openapi/submit, then poll POST /api/openapi/query with the same out_task_id.

Wan 3.0 AI Video Generator and API | 30s, Frames, Docs