GPT-6 Sol & Claude Opus 5.5 are liveGPT-6 Sol at half the price of 5.6 Sol

Wan 3.0 API

AI Video

Wan 3.0 is Alibaba's all-in-one video model on OmniaKey: one id covers text to video, a first frame, and reference-driven shots, and every clip comes back with its own soundtrack.

Alibaba

Wan 3.0

Mode
Prompt0 / 12000
Duration
Resolution
Aspect ratio

Estimated $0.20$0.04/s × 5s

Your result will appear here once you run the model.

01

Where it fits

Wan 3.0 is Alibaba's all-in-one video model on OmniaKey: one id covers text to video, a first frame, and reference-driven shots, and every clip comes back with its own soundtrack.

Choose Wan 3.0 when a shot needs believable audio and steady characters, at $0.04 to $0.16 per second depending on resolution; the same prompt on Wan 3.0 Video Prime lands about twice as fast at a higher rate.

02

What it does

Text to video

wan3.0-video

Generate a short video clip from a text prompt.

Image to video

wan3.0-video

Animate a still image into a short video clip.

Reference to video

wan3.0-video

Guide a clip with reference media to keep a subject, style or motion.

03

In depth

Wan 3.0 is Alibaba's video model on OmniaKey, id wan3.0-video, and the widest single entry in the catalog: text to video, a first frame (optionally with a last frame), and reference-driven shots all run through the same id, at 480P, 720P or 1080P, for up to 30 seconds.

Every clip carries an AAC soundtrack unless the request turns audio off, and Alibaba documents 30 fps output. A task is charged per output second and per resolution, with the charge taken when the task is created and returned if it fails.

On 25 October 2026 we ran three 5-second 720p prompts against Kling 3.0 and Wan took two of them — the noodle stall and the perfume orbit — losing only the skate trick. Those results and a September smoke set, including a 2-second clip, are recorded below.

Shots that need their own sound

Wan writes ambience and effects with the picture, so a market scene gets voices and traffic without a foley pass. Send audio true, or audio false when the clip will be cut to a music track later, and check the track on the first render.

Single takes up to half a minute

Alibaba documents a 2 to 30 second window, which covers a whole product demo in one continuous move. Short jobs are fine too: our smoke set included a 2-second 480P clip, so teasers and stings do not have to be padded.

Perfume-and-product orbits

The label test is the one this model handled best: a bottle, its gold cap and two lines of small type stayed readable through the move on the first frame we supplied. Give it a first frame at shipping size and a slow orbit instruction.

04

Hands-on notes

  • In our three-prompt test on 25 October 2026, Wan 3.0 won two of three clips against Kling 3.0 and lost only the skate action shot.

    Read the test
  • Wan 3.0 finished the same three prompts at a median 144.2 seconds, against 193.6 seconds for Kling 3.0.

    Read the test
  • In our perfume image-to-video test, Wan 3.0 kept the bottle, the gold cap and both lines of label text readable through the camera move.

    Read the test
  • Wan 3.0 also completed a 2-second 480P text-to-video job in our 14 September 2026 smoke run, so the route accepts clips shorter than five seconds.

    Read the test
  • Our 14 September 2026 smoke run returned a 5-second 720P Wan 3.0 text-to-video clip as a 1280x720 MP4 of 5.04 seconds.

05

Limits to plan around

  • 1080P is priced and accepted, but our own smoke set stopped at 720P; run one large-frame job before scheduling a long batch at that size.
  • The output keeps an AAC track even when the request never mentions audio, so a silent master needs audio false plus a check of the downloaded file.
  • Alibaba's own reference still marks Wan 3.0 as a preview model, and 4K is outside what the hosted API offers: 480P, 720P and 1080P are the only published sizes.

06

Compare nearby models

ModelBest forOperationsListed price
Wan 3.0This page
wan3.0-video
Three operations, 30-second takes, 480P to 1080P, sound on by default.Text to video · Image to video · Reference to video
$0.04–$0.16
per second
Kling 3.0
kling-3.0
Fixed 5, 10 or 15 second clips at a charge set before rendering.Text to video · Image to video
$0.80–$3.00
per call
Wan 3.0 Video Prime
wan3.0-video-prime
Same operations and shapes at a higher rate, with a shorter turnaround in our smoke set.Text to video · Image to video · Reference to video
$0.078–$0.38
per second

08

Call it

Media models run asynchronously. Create a task, poll it until it finishes, then download the signed asset URL.

Create, poll, download
# 1. Create the job. 202 Accepted; the body carries the task id,
#    the pre-charged price, and a poll_url. Reuse one Idempotency-Key
#    across retries so a timeout never starts a second paid job.
curl --fail-with-body https://api.omniakey.com/v1/media/tasks \
  -H "Authorization: Bearer $OMNIAKEY_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: wan3.0-video-001" \
  -d '{
    "model": "wan3.0-video",
    "type": "video",
    "operation": "text_to_video",
    "input": {"prompt": "a paper boat on a quiet lake at sunset"},
    "parameters": {"duration": 6, "resolution": "720p"}
  }'

# 2. The same model id and the same poll loop serve the other operations:
#      image_to_video         media[] type "first_frame"
#      reference_to_video     media[] type "reference_image"
#    Reference media travel alongside the prompt:
#    {"prompt": "...", "media": [{"type": "first_frame", "url": "https://example.com/first-frame.png"}]}
#    A public HTTPS URL or a base64 data URI both work.

# 3. Poll GET /v1/media/tasks/{id} until status is
#    succeeded | failed | canceled. Progress carries percent and stage.
curl --fail-with-body https://api.omniakey.com/v1/media/tasks/TASK_ID \
  -H "Authorization: Bearer $OMNIAKEY_API_KEY"

# 4. Download assets[0].url before expires_at, or poll again for a
#    fresh signed URL. Failed and canceled jobs are refunded.
curl -sS -o out.mp4 "$(jq -r '.assets[0].url' result.json)"

09

Pricing

Video jobs are billed per second of rendered output, not per token. Failed or canceled jobs are refunded automatically.

Per second of output

$0.04–$0.16per secondSave 20%

480P
$0.04per second
720P
$0.08per second
1080P
$0.16per second

More models from Alibaba

Related articles

Frequently asked questions

How do I call Wan 3.0 on OmniaKey?

POST to /v1/media/tasks with the model id, type "video", and operation "text_to_video" — or "image_to_video" with a first frame in input.media[] with type "first_frame". Reuse one Idempotency-Key across retries so a timeout never starts a second paid job.

What does Wan 3.0 cost?

Wan 3.0 bills $0.04–$0.16 per second of rendered video. Creation pre-charges the requested duration up front from your prepaid balance, and failed or canceled jobs are refunded automatically.

Why does the API return a task id instead of the video?

Video renders take from about a minute to several minutes, so OmniaKey runs every video route through one asynchronous lifecycle. Poll GET /v1/media/tasks/{id} until the status is succeeded, then read the signed asset URL.

Can I turn off the audio on Wan 3.0?

Yes: send audio false in the parameters. The default is on, and our smoke clips carried an AAC track even where the request had not asked for one, so check the downloaded file rather than the request. The catalog charges one rate per resolution, with or without sound, and current figures are in the price table on the model page.

Does Wan 3.0 really reach 30 seconds?

Alibaba documents a 2 to 30 second output window, and the catalog charges per output second, so a 30-second job is a legitimate request at any resolution. Our own completed jobs ran between 2 and 5 seconds; test one long take at your target resolution before building a schedule around it.

What is the difference between Wan 3.0 and Wan 3.0 Video Prime?

The operations, the 2 to 30 second window and the three resolutions are the same, and Prime costs more at each resolution. In our 720P smoke set the same 5-second text-to-video job finished in 60 to 80 seconds on Prime against 120 to 150 on the standard id.

Is Wan 2.7 video still callable here?

No. The catalog publishes wan3.0-video and wan3.0-video-prime, and the Wan 2.7 video family is not listed; wan2.7-image-pro is an image model and cannot stand in for it. Workflows that depend on Wan 2.7 continuation or instruction editing need a different plan.

Does Wan 3.0 output 4K?

Not through the hosted API. Alibaba's reference lists 480P, 720P and 1080P, and no 4K size appears for either Wan 3.0 id, so a 4K request has nothing to resolve against. Scale up in the edit when a client needs a 4K master.