Skip to main content

Billing for intelligent video splitting

The QuanMiao product matrix includes four Software-as-a-Service (SaaS) products—MiaoBi, MiaoCe, MiaoSou, and MiaoDu—and several AI Agents (scenario-based capabilities). This document describes the billing for the intelligent video splitting application within AI Agents.

Billing overview

We offer real-time and asynchronous modes. We recommend using the asynchronous mode for offline calls.
  • Real-time mode for online calls:
    • Each Alibaba Cloud account receives one free concurrency. You cannot increase the concurrency quota at this time.
    • Supports videos up to 30 minutes in length, less than 200 MB in size, and with a resolution of up to 1080p.
  • Asynchronous mode for offline calls:
    • Each Alibaba Cloud account receives two free concurrencies. You can increase your concurrency quota to a maximum of 30.
    • Supports videos up to one hour in length, less than 450 MB in size, and with a resolution of up to 1080p. The queue can hold up to 100,000 video files.
The free quota for new users of Qwen-series models does not apply to the intelligent video splitting feature.

Billing details

ModePackage BundleBilling
Real-time modeLarge model inference (required)
  • QuanMiao LLM-turbo
    • Input: CNY 0.0008 per 1,000 tokens
    • Output: CNY 0.002 per 1,000 tokens
  • QuanMiao LLM-plus
    • Input: CNY 0.0024 per 1,000 tokens
    • Output: CNY 0.0096 per 1,000 tokens
  • QuanMiao LLM-max (thinking and non-thinking modes)
    • Input: CNY 0.0048 per 1,000 tokens
    • Output: CNY 0.0192 per 1,000 tokens
  • DeepSeek R1
    • Input: CNY 0.004 per 1,000 tokens
    • Output: CNY 0.016 per 1,000 tokens
Multimodal information processing (required)
Includes video format conversion, concatenation, shot detection, frame extraction, and Automatic Speech Recognition (ASR).
CNY 2.8 per hour (billed based on the total duration of successfully processed videos)
Shot enhancement (optional)
Uses Vision Language (VL) and Optical Character Recognition (OCR) models to improve video splitting accuracy.
Select this option based on your video characteristics and splitting results.
  • QuanMiao VL
    • Input: CNY 0.0016 per 1,000 tokens
    • Output: CNY 0.004 per 1,000 tokens
  • OCR
    • Input: CNY 0.0003 per 1,000 tokens
    • Output: CNY 0.0005 per 1,000 tokens
Asynchronous modeLarge model inference (required)
  • QuanMiao LLM-turbo
    • Input: CNY 0.0008 per 1,000 tokens
    • Output: CNY 0.002 per 1,000 tokens
  • QuanMiao LLM-plus
    • Input: CNY 0.0024 per 1,000 tokens
    • Output: CNY 0.0096 per 1,000 tokens
  • QuanMiao LLM-max (thinking and non-thinking modes)
    • Input: CNY 0.0048 per 1,000 tokens
    • Output: CNY 0.0192 per 1,000 tokens
  • DeepSeek R1
    • Input: CNY 0.004 per 1,000 tokens
    • Output: CNY 0.016 per 1,000 tokens
Multimodal information processing (required)
Includes video format conversion, concatenation, shot detection, frame extraction, and Automatic Speech Recognition (ASR).
CNY 2.8 per hour (billed based on the total duration of successfully processed videos)
Shot enhancement (optional)
Uses Vision Language (VL) and Optical Character Recognition (OCR) models to improve video splitting accuracy.
Select this option based on your video characteristics and splitting results.
  • QuanMiao VL
    • Input: CNY 0.0016 per 1,000 tokens
    • Output: CNY 0.004 per 1,000 tokens
  • OCR
    • Input: CNY 0.0003 per 1,000 tokens
    • Output: CNY 0.0005 per 1,000 tokens
Additional concurrency quota (optional)CNY 1 per hour × concurrency count(For example, processing 30 hours of video using a concurrency of 10 costs 1 × 30 × 10 = CNY 300.)
  • Uploaded tasks use the free concurrency quota first. If the free quota is fully utilized, tasks use the paid concurrency quota and are billed based on the actual processing duration of the current task.
  • You can use the GetVideoAnalysisConfig and UpdateVideoAnalysisConfig APIs to view and adjust concurrency.

Activation and billing address

Activation link: You must activate Film & TV Media Video Understanding (free) before you can use intelligent video splitting. Activate video understanding. Billing details: View your bill details. Click the product link to try the product.
  • Technical documentation View the intelligent video splitting API documentation.

FAQ

1. How do I increase concurrency quota for offline calls?

You can use the GetVideoDetectShotConfig - Intelligent video splitting - Get configuration and UpdateVideoDetectShotConfig - Intelligent video splitting - Update configuration APIs to view and adjust concurrency.

2. The charge for a single concurrency is CNY 1 per hour. How is this time defined?

Billing starts when you submit a task and ends when the task is complete.

3. How is billing handled if a task fails midway?

If a task fails during execution, check the final event message. If the event is "task-failed," you are not charged. You are charged only for the portions of the video that were successfully processed. The failure status is immediately displayed on the client.

5. Can I cancel a savings plan for the intelligent video splitting API after purchase?

You can unsubscribe.
Agent Development
Agent Context
Agent Governance & Evolution
Store
Application Marketplace
Use Cases
Support