← Back to all models
Models
omni_flash_8s
Video
Omni video (8s): text · image · first/last frame.
POST/v1/videos
- Modes
- text→video · img→video · first/last frame
- Size
- 16:9 or 9:16 (from size orientation), e.g. 1280x720 / 720x1280
- Seconds
- fixed 8 (set by model name, no need to pass seconds)
- Refs
- optional; 1–2 for first/last frame (first + optional last)
- Billing
- per resolution · per call (tier 720p/1080p)
Model notes
- Duration fixed at 8 seconds by the model name; no need to pass seconds.
- tier sets resolution and billing (720p / 1080p); aspect ratio inferred from size orientation (16:9 / 9:16).
- No reference_images ⇒ text-to-video; with them ⇒ image-to-video.
- First/last frame: set metadata.first_last_frame=true and pass 1–2 reference_images (1st = first frame, 2nd = last frame, order-sensitive); priced as image-to-video.
- Billed per resolution · per call.
Request example
json
{
"model": "omni_flash_8s",
"prompt": "smooth morph between the two frames",
"tier": "1080p",
"size": "1280x720",
"reference_images": ["https://.../first.jpg", "https://.../last.jpg"],
"metadata": { "first_last_frame": true }
}