GENGEN video API
Wan 3.0 Spicy API
Generate a clip from text, first or last keyframes, or multimodal references. Choose a fixed duration or let smart duration select the output length.
wan-3.0-spicySign in to create an API key or submit a generation. A funded wallet is required for paid requests.
When to use Wan 3.0 Spicy
Text-to-video prototypes
Test a scene with a short prompt and fixed duration before adding media references or using smart duration.
Keyframe transitions
Supply a first frame, last frame, or both to guide a transition. Keep keyframe inputs separate from multimodal references.
Reference-based clips
Use image, video and audio references to direct a new clip. Reference-video seconds contribute to both the duration limit and the charge.
Inputs and limits
- The base Spicy model supports 480p, 720p and 1080p. Pro editions have different resolution choices.
- Output is 2–30 seconds, or -1 for smart duration. Reference video totals at most 15 seconds; reference plus output video totals at most 30 seconds.
- Media must be accessible through HTTP(S) URLs without additional headers. Base64 and data URIs are not accepted.
Start with one request
Send this JSON with your bearer API key to POST /v1/contents/generations/tasks.
{
"model": "wan-3.0-spicy",
"mode": "text_to_video",
"prompt": "A sailboat enters a quiet harbor at sunrise. Slow tracking shot.",
"controls": {
"duration": 5,
"resolution": "720p",
"ratio": "16:9"
}
}Save the returned id, then poll the task until it completes. Download the video from outputs.videos.
Wan 3.0 Spicy pricing
The model and resolution rate is multiplied by measured output seconds plus reference-video seconds. Smart duration reserves up to 30 combined seconds.
| Specification | Unit | Price |
|---|---|---|
| 480P output and reference video | second | $0.030004 |
| 720P output and reference video | second | $0.060008 |
| 1080P output and reference video | second | $0.120016 |
Public rates can take up to 15 minutes to refresh. Account-specific prices may differ; your request uses the price applied at submission.
View all current prices →Before your first production request
Start with a short task, save its ID, and verify polling and download handling. Review content errors before retrying; changing models does not guarantee an input will be accepted.