wan-3.0-r2v API

Wan 3.0 offers text-to-video (t2v), first/last-frame image-to-video (i2v) and multi-reference generation (r2v). wan-3.0 runs in China, prime is the faster tier and global is the international version.

Try online · Get an API key · Read the docs

Capabilities

Model
Video
API model
wan-3.0-r2v
Endpoints
POST /v1/video/generations, GET /v1/video/generations/{id}
Parameters
prompt: ≤20000 (t2v / r2v required), seconds: 2–30, metadata.duration: -1 (auto), metadata.resolution: 480P / 720P / 1080P, metadata.ratio: adaptive / 16:9 / 4:3 / 1:1 / 3:4 / 9:16, metadata.generate_audio, metadata.seed: 0–2147483647, images: 1–2 (i2v), ≤10 (r2v), metadata.video_url ≤5, metadata.audio_urls ≤5 (r2v)

Current pricing

≈¥0.90–23.89 per request

Gallery examples

  • Looking Back with PrideLooking Back with Pride

    API model: wan-3.0-r2v

    Looking Back with Pride — explore this wan-3.0-r2v creation on OpenZhenZhen, with its original prompt and a reusable generation example.

Explore the prompt gallery

Quick start

Example workflow: generate a 5-second clip from a prompt with t2v; switch to i2v with 1–2 images to fix the first frame or the first and last frames; use r2v to combine up to 10 images, 5 videos and 5 audio clips, plus one file or web link. Duration is 2–30 seconds, or metadata.duration: -1 for automatic length, and at most 15 seconds with a reference video. i2v and r2v silently replace an out-of-range duration with 5 seconds, so validate it yourself. Resolution is 480P, 720P or 1080P; ratio is adaptive, 16:9, 4:3, 1:1, 3:4 or 9:16. Audio is generated by default (metadata.generate_audio turns it off); seed is 0–2147483647. t2v ignores images.

cURL

curl --fail-with-body https://www.zzai.app/v1/video/generations \
  -H "Authorization: Bearer $SEEDANCE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"images":["https://your-cdn.example/reference.png"],"metadata":{"ratio":"adaptive","resolution":"1080P","video_url":["https://your-cdn.example/reference.mp4"]},"model":"wan-3.0-r2v","prompt":"Create one coherent cinematic shot from all references.","seconds":"5"}'

Python

import json, os, requests

payload = json.loads(r'''{"images":["https://your-cdn.example/reference.png"],"metadata":{"ratio":"adaptive","resolution":"1080P","video_url":["https://your-cdn.example/reference.mp4"]},"model":"wan-3.0-r2v","prompt":"Create one coherent cinematic shot from all references.","seconds":"5"}''')
response = requests.post(
    "https://www.zzai.app/v1/video/generations",
    headers={"Authorization": "Bearer " + os.environ["SEEDANCE_API_KEY"]},
    json=payload, timeout=120,
)
response.raise_for_status()
print(response.json()["id"])

JavaScript

const response = await fetch("https://www.zzai.app/v1/video/generations", {
  method: 'POST',
  headers: { Authorization: 'Bearer ' + process.env.SEEDANCE_API_KEY, 'Content-Type': 'application/json' },
  body: JSON.stringify({"images":["https://your-cdn.example/reference.png"],"metadata":{"ratio":"adaptive","resolution":"1080P","video_url":["https://your-cdn.example/reference.mp4"]},"model":"wan-3.0-r2v","prompt":"Create one coherent cinematic shot from all references.","seconds":"5"}),
});
if (!response.ok) throw new Error(await response.text());
console.log((await response.json()).id);

Save the returned id as TASK_ID in the query example. Query every 3–5 seconds until success or failure; submission does not mean generation is complete. Save result files promptly. data.status: SUCCESS / FAILURE; data.result_url.

curl --fail-with-body https://www.zzai.app/v1/video/generations/$TASK_ID \
  -H "Authorization: Bearer $SEEDANCE_API_KEY"

Frequently asked questions

How do I call this model?

Create an API key, choose the model name below, and send a request to the documented OpenAI-compatible endpoint. Final billing follows the current price and usage parameters.