All Products
Search

This Product

    Platform For AI:iTAG overview

    Document Center

    Platform For AI:iTAG overview

    Last Updated:Jul 06, 2026

    iTAG is PAI's data labeling platform. It includes built-in labeling templates for images, text, video, audio, and multimodal data.

    Important

    Starting August 1, 2026, iTAG 1.0 will switch to allowlist-only access and enter maintenance mode with no further feature updates. Customers who haven't created any labeling tasks or datasets in iTAG 1.0—including new PAI customers—won't be able to use iTAG 1.0. Existing customers who have created labeling tasks or datasets in iTAG 1.0 can continue using the current features. If you need AI-powered automatic dataset labeling, use the Multimodal Data Management feature.

    Supported labeling tasks

    iTAG includes built-in labeling templates that support the following task types:

    • Image: image classification, object detection, image OCR, table recognition, and image semantic segmentation.

    • Text: text classification, named entity recognition, and entity relation recognition.

    • Video: video classification, video timestamping, and video OCR.

    • Audio: audio classification, audio segmentation, and audio recognition.

    • Large model: visual question answering, multimodal RLHF labeling, image-to-text, image-text explanation, dialog rewriting, dialog ranking, and dialog grouping.

    For information about other labeling templates beyond those available in the console, see Manage templates.

    Workflow

    1. Create a dataset

      Upload your data to OSS. Then use the dataset management module to import data from an OSS path and create a dataset. The system generates a .manifest index file (a JSONL file that contains data paths and metadata) for subsequent labeling tasks.

      Important

      iTAG only supports data stored in OSS. For iTAG to access your data, the OSS bucket must be in the same region as PAI.

    2. Create a labeling job

      After creating a dataset, use a built-in or custom template to create and distribute a labeling task. The task distribution workflow includes three phases: labeling, review, and acceptance. Labeling is required, while review and acceptance are optional. Each phase works as follows:

      • Labeling: Annotators claim task packages on the labeling task page, complete the labeling work, and submit the results.

      • Review: Annotators claim completed task packages on the review task page and review, modify, or reject the results.

      • Acceptance: The requester claims task packages on the acceptance task page, performs a final review, and can approve, modify, or reject the results.

    3. Process labeling jobs

      Follow the task workflow to label, review, or accept task packages to produce labeled data.

    4. Export labeling results

      Export the labeling results to a specified OSS directory for model training. The supported export format is .manifest.

    Billing

    • iTAG platform (free): If your team performs manual labeling on iTAG, the platform itself is free.

    • Intelligent labeling service (free): The intelligent labeling service for certain large model labeling templates (such as image-to-text and image-text explanation) is currently free. You'll be notified if pricing changes.

    • Object Storage Service (charged): iTAG depends on Alibaba Cloud Object Storage Service (OSS). Storage and data transfer fees incurred during use are billed separately according to OSS pricing.

    • Outsourced manual labeling service (charged): To have the Alibaba Cloud professional team label your data, submit a ticket or join DingTalk group 21930006619 to contact the PAI team.

    Get help

    If you encounter issues such as data loading errors, missing permissions, or OSS CORS configuration errors, see iTAG FAQ.