Sora 2 Official

Overview

Sora 2 Official API is API Yi’s transparent proxy service for OpenAI’s official API. Unlike reverse-engineered channels, the Official API directly calls OpenAI’s official endpoints, providing higher stability, more precise instruction following, and clearer video quality.

Official Channel: This is a transparent proxy to OpenAI’s official API. All requests are sent directly to OpenAI servers for processing, ensuring the best generation quality and stability.

Official vs Reverse-Engineered Comparison

Comparison	Official API	Reverse-Engineered
Stability	99.99% availability	Subject to risk control, occasional fluctuations
Instruction Following	High precision	Moderate
Video Quality	Consistently clear	Occasionally blurry
Video Duration	4/8/12 seconds	10/15 seconds
Pricing	Per-second billing (higher)	Per-request billing (lower)
Call Method	Async only	Sync streaming / Async

Async Recommended: For both Official and Reverse-Engineered channels, async calls are recommended for better stability and no charges on errors.

Key Features

Direct Official Connection

OpenAI official API transparent proxy, no intermediate processing

Ultimate Stability

99.99% service availability, no risk control issues

High Precision Output

Precise instruction following, consistently clear quality

Flexible Billing

Per-second billing, more economical for on-demand use

Pricing

sora-2 (Standard)

Official model docs: platform.openai.com/docs/models/sora-2

Resolution	Rate	4 sec	8 sec	12 sec
720×1280 (Portrait)	$0.1/sec	$0.4	$0.8	$1.2
1280×720 (Landscape)	$0.1/sec	$0.4	$0.8	$1.2

sora-2-pro (Professional)

Official model docs: platform.openai.com/docs/models/sora-2-pro

Resolution	Rate	4 sec	8 sec	12 sec
720×1280 / 1280×720	$0.3/sec	$1.2	$2.4	$3.6
1024×1792 / 1792×1024	$0.5/sec	$2.0	$4.0	$6.0

Official channel bills by video duration (seconds), not per request
With async calls, failed generations are not charged (e.g., server overload errors)
Videos are stored for 1 day, please download promptly

Requirements

Important: Please complete the following configuration before using the Official channel, otherwise API calls will fail.

Visit api.apiyi.com and log in to your account

Select Group: Sora2 Official

In the API Key Management page, set the group to Sora2官转 (Sora2 Official)

Different groups correspond to different channels and billing methods. Make sure to select the correct one.

Set Billing Mode: Usage-Based

In API Key Settings, set the billing mode to 按量计费 (Usage-Based)

Per-request billing cannot access the Official channel. You must use usage-based billing.

Ensure Sufficient Balance

Official channel pricing is higher, please ensure your account has sufficient balance

API Reference

Endpoint

https://api.apiyi.com/v1/videos

Official API docs: platform.openai.com/docs/api-reference/videos/create

Call Flow (3-Step Async)

The Official channel only supports async calls, requiring three steps:

Submit Request → Get task ID
Poll Status → Check generation progress
Download Video → Get the final video

Step 1: Submit Generation Request

Text-to-Video

curl -X POST "https://api.apiyi.com/v1/videos" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "sora-2",
    "prompt": "A golden retriever running on the beach at sunset, golden sunlight reflecting on the ocean",
    "seconds": "8",
    "size": "1280x720"
  }'

Image-to-Video (Optional)

Use the input_reference parameter to upload a reference image for image-to-video generation.

Image resolution must match the size parameter. For example, if size: "1280x720", the image must also be 1280×720.

curl -X POST "https://api.apiyi.com/v1/videos" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "model=sora-2" \
  -F "prompt=Animate this scene, gentle waves lapping against the sandy beach" \
  -F "seconds=8" \
  -F "size=1280x720" \
  -F "input_reference=@/path/to/image.jpg"

Step 2: Poll Task Status

curl -X GET "https://api.apiyi.com/v1/videos/{task_id}" \
  -H "Authorization: Bearer YOUR_API_KEY"

Step 3: Download Video

import urllib.request

if video_url:
    urllib.request.urlretrieve(video_url, "output_video.mp4")
    print("Video saved to output_video.mp4")

Request Parameters

Parameter	Type	Required	Description
`model`	string	No	Model name: `sora-2` (default) or `sora-2-pro`
`prompt`	string	Yes	Video description prompt
`input_reference`	file	No	Optional reference image for image-to-video. Only 1 file supported, and image resolution must match video resolution
`seconds`	string	No	Video duration, options: `4`/`8`/`12`, default: `4`
`size`	string	No	Output resolution, e.g., `720x1280`, `1280x720`, `1024x1792`, `1792x1024`, default: `720x1280`

Image-to-Video Note: When using input_reference, the reference image resolution must exactly match the video resolution specified in the size parameter, otherwise you’ll get an error: Inpaint image must match the requested width and height

Response Format

Submit Request Response

{
  "id": "video_123",
  "object": "video",
  "model": "sora-2",
  "status": "queued",
  "progress": 0,
  "created_at": 1712697600,
  "size": "1280x720",
  "seconds": "8",
  "quality": "standard"
}

Status Query Response

{
  "id": "video_123",
  "object": "video",
  "model": "sora-2",
  "status": "completed",
  "progress": 100,
  "created_at": 1712697600,
  "size": "1280x720",
  "seconds": "8",
  "quality": "standard",
  "video_url": "https://cdn.example.com/video.mp4"
}

Status Definitions

Status	Description
`queued`	Task submitted, waiting in queue
`in_progress`	Video is being generated
`completed`	Generation complete, ready for download
`failed`	Generation failed

FAQ

What's the difference between Official and Reverse-Engineered?

Official API: Directly calls OpenAI’s official API with high stability, good quality, and precise instruction following, but at a higher price with per-second billing.Reverse-Engineered: Interface implemented through reverse engineering. Lower price with per-request billing, but may be affected by risk control, with slight variations in stability and quality.

Why does Official API only support async calls?

The Official API directly interfaces with OpenAI’s official API, which uses async task mode for video generation. Standard version typically takes 3-5 minutes, while Pro version may take up to 10 minutes depending on complexity. Async mode avoids long connection timeouts and is more stable and reliable.

What durations does Official API support?

The Official API currently supports 4, 8, and 12 second duration options. Unlike the 10/15 seconds of reverse-engineered channels, these are the native durations supported by the official API.

How to get recharge discounts?

API Yi offers various recharge bonus tiers. See Recharge Promotions for details.

Will failed generations be charged?

With async calls, failed video generations are not charged. However, please ensure your input content is compliant and lawful to avoid resource waste and disputes.

How long are generated videos stored?

Videos are stored on CDN for 1 day. Please download them promptly after generation. The video links will expire after this period.

Sora 2 Reverse-Engineered

Lower-priced reverse-engineered channel with per-request billing and sync streaming

Sora 2 Async Guide

Detailed async call tutorial and best practices

Recharge Promotions

View the latest recharge bonus offers

Use Cases

View various client configuration tutorials

Basics

Basic API

Video API

Image API

Multimodal Understanding API

Text API

Overview

Official vs Reverse-Engineered Comparison

Key Features

Direct Official Connection

Ultimate Stability

High Precision Output

Flexible Billing

Pricing

sora-2 (Standard)

sora-2-pro (Professional)

Requirements

API Reference

Endpoint

Call Flow (3-Step Async)

Step 1: Submit Generation Request

Text-to-Video

Image-to-Video (Optional)

Step 2: Poll Task Status

Step 3: Download Video

Request Parameters

Response Format

Submit Request Response

Status Query Response

Status Definitions

FAQ

Sora 2 Reverse-Engineered

Sora 2 Async Guide

Recharge Promotions

Use Cases

Basics

Basic API

Video API

Image API

Multimodal Understanding API

Text API

​Overview

​Official vs Reverse-Engineered Comparison

​Key Features

Direct Official Connection

Ultimate Stability

High Precision Output

Flexible Billing

​Pricing

​sora-2 (Standard)

​sora-2-pro (Professional)

​Requirements

​API Reference

​Endpoint

​Call Flow (3-Step Async)

​Step 1: Submit Generation Request

​Text-to-Video

​Image-to-Video (Optional)

​Step 2: Poll Task Status

​Step 3: Download Video

​Request Parameters

​Response Format

​Submit Request Response

​Status Query Response

​Status Definitions

​FAQ

​Related Resources

Sora 2 Reverse-Engineered

Sora 2 Async Guide

Recharge Promotions

Use Cases

Overview

Official vs Reverse-Engineered Comparison

Key Features

Pricing

sora-2 (Standard)

sora-2-pro (Professional)

Requirements

API Reference

Endpoint

Call Flow (3-Step Async)

Step 1: Submit Generation Request

Text-to-Video

Image-to-Video (Optional)

Step 2: Poll Task Status

Step 3: Download Video

Request Parameters

Response Format

Submit Request Response

Status Query Response

Status Definitions

FAQ

Related Resources