What is Sora2?
Sora2 is an AI video generation model on UniAll AI designed for short-form video creation. It supports two main workflows:
- **Text to video**: generate a video from a written prompt.
- **Reference image to video**: use one image as visual guidance, then animate it with a prompt.
The public model id is `sora2`.
Sora2 is suitable for product videos, ad creatives, social media clips, cinematic concept shots, storyboard previews, and automated video workflows.
Who should use the Sora2 API?
The Sora2 API is useful for:
- Developers building AI video tools or creative automation products.
- Marketing teams generating ad variations and short-form assets.
- E-commerce teams creating product reveal or lifestyle videos.
- Agencies and platforms that need async video generation in their own workflow.
- Businesses that want API-based access without managing model infrastructure directly.
API capabilities
Sora2 on UniAll AI supports async generation through the video generation endpoint.
**Endpoint**
```http POST /v1/videos/generations ```
**Model id**
```json "model": "sora2" ```
**Supported modes**
- `text_to_video`
- `reference_image_to_video`
**Core parameters**
| Parameter | Type | Notes | |---|---:|---| | `prompt` | string | Required. Describes the video to generate. | | `generation_mode` | string | Use `text_to_video` or `reference_image_to_video`. | | `image_url` | string | Required for reference image-to-video mode. Supports PNG, JPEG, or WebP. | | `duration` | integer | Supported options: 4, 8, 12, 16, or 20 seconds. | | `aspect_ratio` | string | `16:9` or `9:16`. | | `resolution` | string | Current request schema exposes `720p`. | | `video_count` | integer | Fixed at 1 video per generation. |
Example request
```json { "model": "sora2", "generation_mode": "reference_image_to_video", "prompt": "A cinematic product reveal, soft studio lighting, smooth camera movement.", "image_url": "https://example.com/reference.png", "duration": 4, "aspect_ratio": "16:9", "resolution": "720p", "video_count": 1 } ```
For text-to-video, omit `image_url` and set:
```json "generation_mode": "text_to_video" ```
Pricing angle
Sora2 is billed by output video duration. The listed user price is **$0.068 per second**. For example, a 4-second generation is estimated from the per-second rate. Failed generations are eligible for refund according to the model configuration, while automatic retry is not enabled by default.
API and partner pricing may differ by account type, usage volume, or commercial arrangement. Check the active UniAll AI pricing profile in your dashboard before production deployment.
Integration notes
Because video generation is async, your application should submit a generation task, store the returned task identifier, and poll or handle task status updates depending on your integration pattern. Use clear prompts, match aspect ratio to the final distribution channel, and keep reference images visually close to the desired output when using image-to-video mode.
UniAll AI provides access to Sora2 through its own platform interface and API layer. This page does not imply an unsupported official partnership or exclusive model provider relationship.
常见问题
How do I call the Sora2 API?
Send a POST request to `/v1/videos/generations` with `model` set to `sora2`. Include a prompt, duration, aspect ratio, resolution, and a generation mode such as `text_to_video` or `reference_image_to_video`.
What generation modes does Sora2 support?
Sora2 supports text-to-video and reference image-to-video. Image-to-video requires a single `image_url` in PNG, JPEG, or WebP format.
How is Sora2 priced on UniAll AI?
Sora2 is billed per second of generated video. The listed user rate is $0.068 per second, with estimated cost depending on the selected duration.