About WAN 3.0 Prime
WAN 3.0 Omni Prime is the speed-optimized tier of the all-in-one reference mode of Alibaba's WAN 3.0 unified video model. Per Alibaba it produces outputs identical in quality to WAN 3.0 Omni while generating 5 to 7 times faster, at a higher price per second; a 15-second 720P clip takes roughly two minutes. It generates 2 to 30 second clips at up to 1080P, with generated audio on by default, and supports a smart duration mode that lets the model pick the length from the prompt. With no references supplied it generates from the prompt alone.
It accepts up to 10 images, 5 video clips, and 5 audio clips as combined references for appearance, motion, and voice.
It also supports first/last frame generation as an alternative mode; frame inputs and reference inputs cannot be combined in a single request.
See what this model can do
A handful of real prompts and their outputs.
Ready to build with WAN 3.0 Prime?
Try WAN 3.0 Prime in the Workbench to prompt it, compare outputs, and iterate on prompts without writing any code. When you're ready to ship, call the same model from our API and build your own apps on top of it.
curl -sSf -X POST https://hub.oxen.ai/api/ai/videos/generate \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $OXEN_API_KEY" \
-d '{
"model": "wan-v3-0-video-prime",
"prompt": "A kitten running under the moonlight, moonlight falling on its soft fur, with a serene garden in the background",
"generate_audio": true,
"aspect_ratio": "adaptive",
"resolution": "1080P",
"duration": 5,
"enable_thinking": false,
"watermark": false
}'Request parameters
Every field WAN 3.0 Prime accepts in the request body. See the API reference for response formats.
https://hub.oxen.ai/api/ai/videos/generate| Field | Type | Required | Default | Description |
|---|---|---|---|---|
prompt | string | Yes | — | Text prompt describing the desired video. Refer to reference assets with the @ syntax: @Image 1, @Image 2, ..., @Video 1, ..., @Audio 1, ...; each type is numbered separately. |
input_images | array<string>nullable | No | — | Reference images for character, object, or scene appearance. Order maps to @Image 1, @Image 2, etc. in the prompt. Up to 10 images, max 20 MB each. JPEG/PNG/BMP/WEBP. |
input_image | stringuri | No | — | Optional first frame image the video is generated from. JPEG/PNG/BMP/WEBP, max 20 MB. The output aspect ratio is taken from this image. Cannot be combined with reference images, videos, or audio. |
tail_image_url | stringuri | No | — | Optional last frame image. The video interpolates from the first frame to this one. Requires a first frame image; same format constraints. Cannot be combined with reference images, videos, or audio. |
input_videos | array<string>nullable | No | — | Reference videos for character or object appearance, motion, voice, or a camera movement to follow. Order maps to @Video 1, @Video 2, etc. in the prompt. Up to 5 clips totalling 15 seconds, max 100 MB each. MP4/MOV. |
input_audios | array<string>nullable | No | — | Reference audio used as a voice or sound reference. Order maps to @Audio 1, @Audio 2, etc. in the prompt, e.g. "using the voice from @Audio 1". Up to 5 clips totalling 15 seconds, max 15 MB each. WAV/MP3. |
generate_audio | boolean | No | true | Whether the model generates sound for the video. Pricing is the same either way. |
aspect_ratio | string | No | "adaptive" | adaptive16:99:161:14:33:4 |
resolution | string | No | "1080P" | 480P720P1080P |
duration | integer | No | 5 | Range: 2 – 30 |
enable_thinking | boolean | No | false | Enables thinking mode, which lets the model reason about the prompt before generating. Adds processing time. |
seed | integer | No | — | Random seed for reproducibility (0-2147483647). |
watermark | boolean | No | false | Adds an 'AI generated' watermark to the bottom-right corner. |