Multimodal Content Moderation Service

Use cases

Alibaba Cloud Content Moderation provides a multimodal moderation service powered by a large model to analyze content that combines images and text. The service more effectively detects risk content by examining images, associated text, and their context. The following use cases are supported:

• Analyze posts and comments: In community forums and discussion boards, the service comprehensively assesses risks by analyzing the main post and its associated comments. A main post includes a title, body text, and images. Comments include text and images, and support nested replies. The service evaluates all these elements together to detect risk content.

• Analyze user profiles: For applications such as social media, instant messaging (IM), and gaming, the service performs a comprehensive risk analysis on user profiles. It combines a user's avatar (image) and nickname (text) to determine if they constitute risk content.

Services

The multimodal content moderation feature supports the following services:

Billing

The AIGC detection service for Image-Text Hybrid Moderation Enhanced Edition supports two billing methods: pay-as-you-go and resource package deduction.

Pay-as-you-go

After you activate the Image-Text Hybrid Moderation Enhanced Edition service, pay-as-you-go is the default billing method. You are charged daily based on your actual usage. You incur no fees if you do not use the service.

Getting started

1. Sign up for an Alibaba Cloud account. If you do not have an account, register now.

2. To activate pay-as-you-go for Content Moderation, make sure that you have activated the service. For more information, see Activate Service. Activation is free of charge. After you start to use the API, the system automatically generates bills based on your usage. For more information, see Billing (to be determined). You can also purchase pay-as-you-go resource packages. Compared to the standard postpaid method, resource packages offer tiered discounts and are suitable for users with predictable and high usage. For more information, see Purchase Pay-as-you-go Resource Package.

3. Create an AccessKey: Ensure that you have created an AccessKey by using RAM. For more information, see Create an AccessKey. If you are using the AccessKey of a RAM user (sub-account), you must use an Alibaba Cloud account (main account) to grant the AliyunYundunGreenWebFullAccess permission to the RAM user. For more information, see RAM Authorization.

4. Integrate the service by using an SDK, the recommended method for making API calls. See the Multimodal Content Moderation SDK and Integration Guide.

Submit a moderation task

API

• API operation: MultimodalAsyncModeration.

• Supported regions and endpoints:

• Billing: This is a paid API. You are charged based on the number of submitted images and the character count of the text. For more information, see Billing.

• Detection object: The service detects text and image content.

• Response: This is an asynchronous operation. Results are not returned in real time. You can retrieve them through a callback or by polling. The results are retained for up to 24 hours.

  • Callback: To automatically receive detection results, specify the callback parameter in your request.

  • Polling: If you do not specify the callback parameter, you must call the result query API to retrieve the detection results.

• Data Requirements:

  • The total text length cannot exceed 5,000 characters, and the total number of images cannot exceed 30.

  • Supported image formats: PNG, JPG, JPEG, BMP, WEBP, TIFF, SVG, HEIC (the longest side of an HEIC image must be less than 8,192 pixels), GIF (only the first frame is processed), and ICO (only the last image is processed).

  • The image size cannot exceed 20 MB. The height or width cannot exceed 16,384 pixels, and the total number of pixels cannot exceed 167 million. We recommend a minimum resolution of 200 x 200 pixels. Low resolution may reduce moderation accuracy.

  • The image download is limited to 3 seconds. A download that exceeds this limit returns a timeout error.

Request parameters

Table 1. ServiceParameters

Table 2. mainData

Table 3. mainImages

Table 4. commentDatas

Table 5. images

Table 6. commentDatas

{
    "service": "post_text_image_detection",
    "serviceParameters": {
        "dataId": "Multimodal0424***",
        "mainData": {
            "mainTitle": "This is a title or user profile",
            "mainContent": "This is the main text or user nickname",
            "mainImages": [
                {
                    "imageUrl": "https://aliyun.com/240308/test001.jpg"
                },
                {
                    "imageUrl": "https://aliyun.com/240308/test002.jpg"
                }
            ],
            "mainPostTime": "2025-06-18 20:20:20"
        },
        "commentDatas": [
            {
                "images": [
                    {
                        "imageUrl": "https://aliyun.com/240308/test003.jpg"
                    },
                    {
                        "imageUrl": "https://aliyun.com/240308/test004.jpg"
                    }
                ],
                "content": "Comment 1",
                "postTime": "",
                "commentDatas": [
                    {
                        "content": "Reply 1 to Comment 1",
                        "images": [
                            {
                                "imageUrl": "https://aliyun.com/240308/test005.jpg"
                            }
                        ],
                        "postTime": ""
                    },
                    {
                        "content": "Reply 2 to Comment 1",
                        "images": [
                            {
                                "imageUrl": "https://aliyun.com/240308/test006.jpg"
                            }
                        ],
                        "postTime": ""
                    }
                ]
            },
            {
                "content": "Comment 2",
                "images": [],
                "postTime": "",
                "commentDatas": []
            },
            {
                "content": "Comment 3",
                "images": [],
                "postTime": "",
                "commentDatas": []
            },
            {
                "content": "Comment 4",
                "images": [],
                "postTime": "",
                "commentDatas": []
            },
            {
                "content": "Comment 5",
                "images": [],
                "postTime": "",
                "commentDatas": []
            },
            {
                "content": "Comment 6",
                "images": [],
                "postTime": "",
                "commentDatas": []
            },
            {
                "content": "Comment 7",
                "images": [],
                "postTime": "",
                "commentDatas": []
            }
        ]
    }
}

{
    "Code": 200,
    "Data": {
        "DataId": "Multimodal0424***",
        "TaskId": "task_1713942548822_1t4qI8"
    },
    "Message": "OK",
    "RequestId": "AAAAA-BBBBB-AIXI-1314-CCCCC"
}

Retrieve multimodal moderation results

Request parameters

Response parameters

Table 2. Data

Table 3. MainData

Table 4. Result

Table 5. CommentDatas

Table 6. CommentDatas (nested)

Table 7. Usage

Examples

Sample request

{
    "ReqId": "AAAAA-BBBBB-AIXI-1314-CCCCC"
}

Sample responses

• Non-compliant content detected

{
    "RequestId": "ABCD1234-1234-1234-1234-123****",
    "Message": "OK",
    "Code": 200,
    "Data": {
        "DataId": "dataId***",
        "RiskLevel": "high",
        "MainData": {
            "Result": [
                {
                    "Label": "violent_explosion",
                    "Description": "Content may contain elements related to fireworks."
                },
                {
                    "Label": "sexual_partialNudity",
                    "Description": "Suggestive content"
                }
            ]
        },
        "CommentDatas": [
            {
                "Result": [
                    {
                        "Label": "violent_explosion",
                        "Description": "Content may contain elements related to fireworks."
                    },
                    {
                        "Label": "sexual_partialNudity",
                        "Description": "Suggestive content"
                    }
                ],
                "CommentDatas": [
                    {
                        "Result": [
                            {
                                "Label": "violent_explosion",
                                "Description": "Content may contain elements related to fireworks."
                            },
                            {
                                "Label": "violent_explosion",
                                "Description": "Content may contain elements related to fireworks."
                            }
                        ]
                    },
                    {
                        "Result": [
                            {
                                "Label": "nonLabel",
                                "Description": "No non-compliant content detected"
                            }
                        ]
                    }
                ]
            },
            {
                "Result": [
                    {
                        "Label": "violent_explosion",
                        "Description": "Content may contain elements related to fireworks."
                    }
                ]
            },
            {
                "Result": [
                    {
                        "Label": "nonLabel",
                        "Description": "No non-compliant content detected"
                    }
                ]
            }
        ],
        "Usage": {
            "Total_tokens": 1039,
            "Output_tokens": 56,
            "Input_tokens": 983
        }
    }
}

• No non-compliant content detected

{
    "RequestId": "ABCD1234-1234-1234-1234-123****",
    "Message": "OK",
    "Code": 200,
    "Data": {
        "DataId": "dataId***",
        "RiskLevel": "none",
        "MainData": {
            "Result": [
                {
                    "Label": "nonLabel",
                    "Description": "No non-compliant content detected"
                }
            ]
        },
        "CommentDatas": [
            {
                "Result": [
                    {
                        "Label": "nonLabel",
                        "Description": "No non-compliant content detected"
                    }
                ],
                "CommentDatas": [
                    {
                        "Result": [
                            {
                                "Label": "nonLabel",
                                "Description": "No non-compliant content detected"
                            }
                        ]
                    },
                    {
                        "Result": [
                            {
                                "Label": "nonLabel",
                                "Description": "No non-compliant content detected"
                            }
                        ]
                    }
                ]
            },
            {
                "Result": [
                    {
                        "Label": "nonLabel",
                        "Description": "No non-compliant content detected"
                    }
                ]
            },
            {
                "Result": [
                    {
                        "Label": "nonLabel",
                        "Description": "No non-compliant content detected"
                    }
                ]
            }
        ],
        "Usage": {
            "Total_tokens": 1039,
            "Output_tokens": 56,
            "Input_tokens": 983
        }
    }
}

Code

This topic describes the code values returned by the multimodal content moderation API. Metering and billing apply only to requests that return a code of 200 or 280. You are not charged for requests with other code values.