AlibabaAlibaba video model

Wan 3.0. Hold one shot for up to 30 seconds.

Alibaba's Wan 3.0 renders a single continuous take anywhere from 2 to 30 seconds, with native sound, from a prompt or a still. Ask for it by name in Treza chat, or pick Wan 3.0 Prime for the faster mode.

  • 2 to 30 second clips
  • 5 aspect ratios
  • Native audio
  • First frame control
  • Up to 1080p
  • Wan 3.0 Prime fast mode

Made with Wan 3.0

Two briefs, one model. Each frame is a single Wan 3.0 generation, looping silently here.

An overhead push-in on a plated scallop dish with steam curling upward
A slow overhead push-in on a plated dish, steam and all.
A narrator at a studio microphone in a dark acoustically treated booth
A narrator at a studio microphone, on a slow push-in.

Every frame above was generated on Treza withWan 3.0

Wan 3.0 is the long-take model in the Wan family. Where Wan 2.7 stops at 10 seconds, Wan 3.0 runs to 30 in one generation, and you choose the exact second, so an establishing shot or a narrated beat plays out without a cut. It renders in five aspect ratios, generates ambience and effects with the picture, and can start from a still you hand it.

It comes in two modes that share one spec sheet. Wan 3.0 is the standard mode at the lower rate per second; Wan 3.0 Prime is the fast-mode variant for when turnaround matters more than price. In Treza chat, just ask for Wan 3 or Wan 3.0 Prime. On the canvas, both sit in the Video Generation node's model picker, one dropdown apart.

Up to 30 seconds, to the second

Any exact length from 2 to 30 seconds. A shot can match a narration line or a music cue instead of being cut down to a fixed step.

Sound with the take

Ambience and effects are generated with the picture across the whole take. Switch audio off per generation when a narrator or music bed will sit on top.

Start from a still

Hand it an image as the exact first frame and Wan 3.0 animates from there, so a product photo or a generated still keeps its composition.

Standard or Prime

Wan 3.0 Prime keeps the same durations, formats, and audio in a faster mode at a higher per-second rate. The prompt carries across unchanged.

What to make with Wan 3.0

Real prompts, and what comes back. Every one runs in Treza chat or as a node on the canvas.

“Food commercial: overhead slow push-in on a plated seared scallop dish, steam curling upward, warm restaurant light”

A premium food beat with steam and sauce catching the light.

“A 30 second drone glide along a sea cliff at golden hour, one continuous take, 16:9”

One uncut establishing shot, no stitching.

“Animate this product still (attached): slow orbit, light sweeping across the surface”

Image-to-video that keeps your framing from the first frame.

“A narrator at a studio microphone in a dark booth, warm rim light, slow push-in, 22 seconds”

A b-roll take sized to a 22 second voiceover.

“A 2 second insert: a match strikes and flares in the dark”

A short cutaway for an edit, from the bottom of the range.

“The same cliff shot on Wan 3.0 Prime, 9:16 for Shorts”

The fast mode on an identical brief, in vertical.

Wan 3.0 is one node on a canvas

Chain it with language, image, and video models, then publish the whole pipeline as an API. See how a full AI video pipeline fits together.

  1. Step 01

    Draft the brief

    A language model node expands a one-line idea into a detailed, on-brand prompt before it ever reaches the model.

  2. Step 02

    Generate with Wan 3.0

    Run variants in parallel, compare against other models on the same prompt, and keep every result in run history.

  3. Step 03

    Publish it as an API

    The pipeline becomes a versioned endpoint. Your product calls it with the typed /invoke API or any OpenAI SDK.

Wan 3.0 as an API, priced per clip

The Wan 3.0 API

Build a pipeline with Wan 3.0 on a node and publish it. It becomes a versioned HTTP endpoint: the typed /invoke route enqueues a run and returns a run id to poll, and an OpenAI-compatible route works with SDKs you already use. Agents connect over MCP the same way. Your prompt engineering ships inside the pipeline, so every caller gets your quality, and swapping the model later does not change the API your product calls.

Wan 3.0 pricing

Prepaid credits, no subscription. Each clip is charged at Wan 3.0's own rate from your balance, shown before you run, and only successful generations are charged, so a failed run costs nothing. Credit packs start at $5 and never expire, which suits spiky production schedules. Run history records the cost of every generation, per node, so spend is auditable rather than a surprise.

Wan 3.0, answered

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

They take the same inputs: 2 to 30 second clips, five aspect ratios, native audio, and an optional first frame. Prime is the fast-mode variant and bills a higher rate per second. Start on Wan 3.0 and switch to Prime when turnaround matters more than price. In chat, name either one; on the canvas, it is one dropdown on the node.

How is Wan 3.0 different from Wan 2.7?

Wan 3.0 holds a single take three times as long, up to 30 seconds against 10, with the same five aspect ratios and native audio. Wan 2.7 keeps two things of its own: reference images, including photos of real people, and a last frame keyframe. Reach for Wan 3.0 for long takes and Wan 2.7 for people-led clips built from photos. Both are in Treza chat and in the same picker.

How does Wan 3.0 compare to Seedance 2.5 for 30 second clips?

Both render any exact length up to 30 seconds with native audio. Wan 3.0 starts as short as 2 seconds, where Seedance 2.5 starts at 4. Seedance 2.5 adds a last frame keyframe and 21:9 ultrawide, so it is the pick when a take has to land on an exact closing frame. For everything else, run the brief on both, one dropdown apart, and keep the better take.

How much does Wan 3.0 cost on Treza?

Treza uses prepaid credits with no subscription. Each clip is charged at Wan 3.0's rate from your balance, and only successful generations are charged. Credit packs start at $5 and never expire.

Can I call Wan 3.0 through an API?

Yes. Build a pipeline that uses Wan 3.0 and publish it as a versioned HTTP endpoint. Call the typed /invoke API from your product, or point any OpenAI SDK at the OpenAI-compatible endpoint. If a better model ships later, swap it on the node without changing the API your product calls.