AI Video APIA genuinely unified public contract
Applications switch among text, frame control, references, documents, and web context by changing standard input fields, while retaining one model ID, task endpoint, callback format, and billing account.
All-in-One multimodal video API
Wan 3.0 API
One Wan 3.0 Video API for text, frames, rich media, files, and web pages.
Wan 3.0 All-in-One unifies text-to-video, first/last-frame animation, multimodal reference generation, and content understanding for documents and public web pages. AI Video API exposes the invited Beijing Workspace model through one asynchronous endpoint, exact pre-charging, callbacks, and a consistent task response.
Read the official model and pricing documentation2–30s
Output duration
480p / 720p / 1080p
Resolution
Image + video + audio + file + web
Reference inputs
Alibaba
wan-3.0-videoasync task
2–30s
Output duration
480p / 720p / 1080p
Resolution
Image + video + audio + file + web
Reference inputs
Model capabilities
Built for production API workflows, not one-off demos.
Use Wan 3.0 from the same platform surface as the rest of your video and image stack: API keys, credits, logs, webhooks, and docs stay consistent across providers.
One model for every generation mode
Use the same wan-3.0-video model ID for text-only generation, a strict first frame, first and last frames, or rich multimodal references. Mode selection follows the supplied media instead of requiring separate public model names.
Ordered multimodal references
Attach up to 10 images, five videos, and five audio files. Wan 3.0 preserves each media array's order so prompts can refer to Image 1, Video 1, or Audio 1 while combining visual identity, motion, and sound intent.
File and public web understanding
Use a supported public document URL or a login-free web page as source material. The model can interpret the content and turn it into a video brief without first converting every page into image references.
Up to 30 seconds with audio
Choose an integer duration from 2 to 30 seconds and keep generated audio enabled by default. When reference video is used, its detected total plus output duration must remain within the official 30-second limit.
Billing-safe asynchronous workflow
The API probes every reference video and audio file before charging. It pre-charges for the requested output plus detected input-video duration, refunds confirmed failures, and never refunds a successful generation.
API workflow
Submit tasks, track progress, and return generated assets.
01
Choose text, frames, or references
Write a prompt up to 20,000 characters or provide public HTTP(S) media. Frame mode stays separate from reference media, files, and web links, while file_url and link_url remain mutually exclusive.
02
Select duration, ratio, and resolution
Choose 2–30 seconds, adaptive or a supported fixed ratio, and 480p, 720p, or 1080p. Auto duration (-1) is intentionally rejected so the platform can pre-charge accurately.
03
Submit the task
Submit the generation request directly. The platform detects reference-video duration before charging and applies the selected resolution rate.
04
Poll or receive callbacks
Track the task through GET /v1/tasks/{taskId} or a callback URL. Wan 3.0 tasks run to a completed or failed terminal state and do not expose task cancellation.
POST /v1/videos/generations
curl -X POST https://api.aivideoapi.ai/v1/videos/generations -H "Authorization: Bearer sk-your-api-key" -H "Content-Type: application/json" -d '{
"model": "wan-3.0-video",
"callback_url": "https://your-server.com/webhook",
"input": {
"prompt": "Video 1 provides the camera move, Image 1 provides the product identity, and Audio 1 guides the soundtrack",
"image_urls": ["https://example.com/product.png"],
"video_urls": ["https://example.com/camera.mp4"],
"audio_urls": ["https://example.com/music.mp3"],
"resolution": "720p",
"aspect_ratio": "adaptive",
"duration": 5,
"audio": true
}
}'What teams build
Text-to-video campaigns
First and last frame transitions
Character and product reference videos
Motion and camera reference transfer
Audio-guided storytelling
Document-to-video explainers
Public article summaries
Longer 30-second creative sequences
Why choose Wan 3.0
The Wan 3.0 API benefits that matter to developers.
Exact reference-video billing
Server-side duration detection runs during generation submission. A 4.5-second reference plus five seconds of 720p output produces a 228-credit charge.
Explicit operational boundaries
The public contract accepts only public HTTP(S) URLs, enforces individual and combined media limits, and documents temporary 24-hour provider links so production systems know when to download or enable storage mirroring.
Frequently asked questions
Answers about the Wan 3.0 API.
What is the Wan 3.0 API?
The Wan 3.0 API is AI Video API's integration for Alibaba's Wan 3.0 All-in-One video model. Use POST /v1/videos/generations with model wan-3.0-video for text, first/last frames, multimodal references, documents, or public web pages, then receive an asynchronous task ID.
Which resolutions does the Wan 3.0 Video API support?
The documented tiers are 480p, 720p, and 1080p. The default is 1080p. This integration does not claim native 4K support because Alibaba's current Wan 3.0 API documentation lists only those three resolution tiers.
How much does Wan 3.0 cost?
Wan 3.0 costs 12 credits per second at 480p, 24 at 720p, and 48 at 1080p. Reference-video requests bill detected input video seconds plus requested output seconds. Images, audio, files, and web pages do not add input-video seconds.
What are the Wan 3.0 media limits?
A request supports up to 10 reference images, five reference videos, and five reference audio files. Each video or audio item must be 1–15 seconds and each category can total at most 15 seconds. Input video plus output video cannot exceed 30 seconds.
Can I mix first/last frames with reference media?
No. First/last-frame mode is mutually exclusive with reference images, videos, audio, file_url, and link_url. A last frame also requires a first frame, and file_url cannot be combined with link_url.
How long is the Wan 3.0 result URL available?
Alibaba's upstream result link is available for about 24 hours. Accounts with Bailian R2 mirroring enabled receive a stored copy; otherwise download the result promptly after the task completes.
Pricing and usage
Clear model options with shared credits.
480p
12 credits/s
60 credits for 5 seconds without input video
720p
24 credits/s
120 credits for 5 seconds without input video
1080p
48 credits/s
240 credits for 5 seconds; default tier
Reference video
Input + output seconds
Both use the selected resolution rate
Start building with Wan 3.0 in AI Video API.
Create one API key, use one credit balance, and switch between video and image models without provider-specific plumbing.