Quanmiao Lite App release notes

Date

Feature module

Change

Description

November 28

General Enterprise VOC Mining

Template expansion

Adds four new scenario templates to enrich the existing eight templates: low-value fast-moving consumer goods (FMCG) e-commerce feedback, bank/insurance customer service experience analysis, cultural and entertainment industry (such as novels and movies) review analysis, and restaurant user review analysis. Go to the General Enterprise VOC Mining console to try them.

Date

Feature module

Change

Description

October 27

Film and Media Video Understanding

Increased concurrency limit

The concurrency limit can be increased to a maximum of 30.

New vision model: Qwen-VL-Post

Qwen-VL-Post provides the capability to describe film and media works with faster inference speed and lower costs. Go to the Film and Media Video Understanding console to try it.

October 24

Smart Video Splitting

Product release

Smart Video Splitting is officially released. This feature splits videos, such as news and programs, into segments based on their content. It also outputs the timeline, title, and summary for each segment. Go to the Smart Video Splitting console to call the API and try it.

Date

Feature module

Change

Description

September 10

In-vehicle Infotainment Hotspot Q&A

Timeliness sorting optimization

Increases the weight of timeliness when returning the news recommendation list, which improves the real-time delivery of news.

Date

Feature module

Change

Description

August 22

Film and Media Video Understanding

Language support expansion

The number of supported languages is expanded from 4 to 12. The newly added languages are Arabic, Korean, Malay, Thai, Portuguese, Spanish, Indonesian, and Vietnamese.

August 13

Sample prompts

Adds prompts for the "secondary processing" and "video content quality inspection" scenarios.

Date

Feature module

Change

Description

July 31

One-click Video Generation

Feature update

• Viral video imitation: Create viral marketing video clips by imitating the style of voice-over scripts.

• Long-to-short video editing: The granularity of the video understanding step is refined to 1 second. Supports custom voice-over scripts. Expands text-to-speech (TTS) system voices. Adds a video material transfer feature to save materials in Object Storage Service (OSS). Refines the editing timeline, with clip precision to the millisecond and removal of unwanted shots.

• The total length of supported source materials is increased to 20 minutes.

July 31

Film and Media Video Understanding

OCR character recognition

Adds a Platform as a Service (PaaS) API that uses Optical Character Recognition (OCR) to automatically identify on-screen characters. This allows for automatic identification of character information in films and TV shows.

July 4

Essay Grading Assistant

New essay grading model

Lets users select different grading output structures based on their preferences.

Film and Media Video Understanding

SRT format support

In the "Automatic speech recognition settings" option, custom Automatic Speech Recognition (ASR) files now support the SRT format.

Date

Feature module

Change

Description

June 27

Essay Grading Assistant

Lite app launched

Automatically analyzes student essays to check for grammar, spelling errors, and classical allusions. It quickly generates structured scores, comments, and revision suggestions to help teachers grade efficiently.

Try it out: Essay Grading Assistant console

For more information, see Essay Grading Assistant.

June 6

Quanmiao - Hotspot News Broadcast MCP Server

Launched

Hotspot News Broadcast MCP Server URL

Date

Feature module

Change

Description

May 30

General Enterprise VOC Mining

Label traceability support

Outputs the original VOC text and position for each label.

E-commerce and Retail Promotional Copywriting

Support for more models

Supports more models, including thinking models.

May 23

Film and Media Video Understanding

Text processing supports Qwen3.

Custom ASR is supported.

• Adds the Qwen3-235B-A22B model, which includes both thinking and non-thinking modes.

• For more information about billing, see Billing of Film and Media Video Understanding.

• For more information about custom ASR, see SubmitVideoAnalysisTask - Submit an asynchronous video understanding task.

Media/Retail Article Style and Format Learning

Upload local documents for sample articles

Upload local documents as sample articles. Supported formats are PDF and TXT. The maximum file size is 20 MB.

May 16

• Film and Entertainment Screenwriting

• Media/Retail Article Style and Format Learning

• Film and Media Video Understanding

• Film and Entertainment Screenwriting and Media/Retail Article Style and Format Learning are commercially available.

• For Film and Media Video Understanding, online and offline PaaS APIs now support custom ASR. A new PaaS API is added to cancel tasks that have not started scheduling.

• For more information about billing, see the following documents:

  • Billing of Film and Entertainment Screenwriting

  • Billing of Media/Retail Article Style and Format Learning

• Custom ASR APIs:

  • Online: RunVideoAnalysis - Run an online video understanding task

  • Offline: SubmitVideoAnalysisTask - Submit an asynchronous video understanding task

• API to cancel unscheduled tasks: UpdateVideoAnalysisTask - Update the status of an asynchronous video understanding task

Date

Feature module

Change

Description

April 25

In-vehicle Infotainment Hotspot Q&A

• Integrated with Cover News

• MCP Server is supported

Integration with Cover News adds support for politics, economy, military, and international channels.

April 18

• Online Content Moderation

• General Enterprise Lead Mining

• E-commerce and Retail Promotional Copywriting

Commercially available

For more information about billing, see the following documents:

April 16

General Enterprise VOC Mining lite app

Debugging interface redesign

• Simplified the material upload process.

• Optimized sample filling.

For more information, see General Enterprise VOC Mining.

April 11

General Enterprise VOC Mining

Commercially available

For more information about billing, see Billing of General Enterprise VOC Mining.

April 1

• Film and Media Video Understanding

• In-vehicle Infotainment Hotspot Q&A

• Film and Entertainment Screenwriting update

Improved stability and feature updates

• Film and Media Video Understanding

  • Uses Function Compute (FC) elastic scaling to improve throughput and stability.

  • Supports calls from other workspaces under the same account.

  • The validity period of VL files is adjusted to 4 hours to prevent tasks for long videos from expiring after 1 hour.

  • The API lets you disable ASR. This makes the Agent ignore audio information and only recognize and understand information from sampled video frames.

• In-vehicle Infotainment Hotspot Q&A

  • Optimized the clustering algorithm policy for news hotspots.

  • The update frequency for the official broadcast list on the news hotspot API page is changed from six times a day to three times a day.

  • The hotspot calculation order can no longer be customized.

  • Adds a personalized content recommendation feature.

• Film and Entertainment Screenwriting update

  The maximum character limit for screenplay input is expanded to 8,000 characters.

Date

Feature module

Change

Description

March 14

Film and Media Video Understanding

• New sample templates

• New Deepseek-r1 model

• Adds three sample templates to meet VL analysis requirements in different scenarios.

• Adds the Deepseek-r1 model to the list of supported Large Language Models (LLMs).

Date

Feature module

Change

Description

February 21

Film and Media Video Understanding

• New model for text processing

• Optimized token billing model

• Increased character limit for supplementary text

• Adds the Qwen2.5-7B-1M model for text processing.

• When you call Qwen-Max, the prices for input/output tokens are aligned with the latest base model prices.

• The character limit for supplementary text related to videos is increased from 400 to 600 characters.

For more information, see Film and Media Video Understanding.

February 18

Film and Media Video Understanding

Expanded support for uploaded video formats

Supports videos in .flv, .ts, .avi, .wmv, .mpg, .mkv, .mov, and .mp4 formats. The video size must be less than 150 MB, the resolution must be less than 1080p, and the duration must not exceed 60 minutes.

For usage instructions, see Film and Media Video Understanding.

Date

Feature module

Change

Description

January 24

• Online Content Moderation

• General Enterprise Lead Mining

• Media/Retail Article Style and Format Learning

• Online Content Moderation

  • New "Article Proofreading Scenario" sample

  • New batch processing API

• General Enterprise Lead Mining

  • New scenario samples

  • New feature to add content tags in batches

  • Optimized format for lead mining results

  • New batch processing API

• Media/Retail Article Style and Format Learning

  Optimized user experience and upgraded the learning and writing workflow

• Online Content Moderation

  • Adds an "Article Proofreading Scenario" sample to help users check materials for issues such as typos, incorrect tags, and awkward or ungrammatical sentences.

  • Adds a batch processing API.

• General Enterprise Lead Mining

  • Adds scenario samples for "consumer behavior analysis for furniture purchases" and "corporate partnership information mining".

  • Adds a feature to add content tags in batches.

  • Optimizes the output format of lead mining results. You can copy the generated results and paste them directly into an Excel sheet.

  • Adds a batch processing API.

• Media/Retail Article Style and Format Learning

  Improves the first-packet response time and significantly reduces the number of tokens consumed by the large model during content creation.