What is Platform for AI? - Platform For AI - Alibaba Cloud Documentation Center

What is PAI

Platform for AI (PAI) is a one-stop machine learning platform for developers by Alibaba Cloud. With its modules, It provides end-to-end AI development services including data labeling, model development, training, and deployment.

Module	Description	Scenario
iTAG	Supports data labeling for different types of data such as images, text, and videos. It also provides fully managed outsourced data labeling services.	Data labeling
Data Science Workshop (DSW)	A cloud-based AI development IDE or machine with multiple built-in environments. Quickly start model development if you are familiar with Notebook or VSCode.	Model development
Machine Learning Designer	Has 140+ built-in algorithm components. It allows low-code model building through visualized drag-and-drop operations.	Big data + Model development
Deep Learning Containers (DLC)	Easily creates distributed or standalone training jobs without changing your habits. Its underlying infrastructure is based on Kubernetes, eliminating the need to manually purchase machines and configure runtime environments.	Distributed model training
Elastic Algorithm Service (EAS)	With simple configuration, deploys trained models as online services for use in other applications.	Deploy models as service interfaces
Model Gallery	Especifically designed for large language models (LLMs). It encapsulates PAI-DLC and PAI-EAS, helping you train and deploy LLMs efficiently and quickly with zero code.	Zero-code training and deployment of open-source models

Benefits

End-to-end AI-powered R&D

Supports data labeling, model development, model training, model optimization, model deployment, and AI O&M as a one-stop AI platform.
Provides over 140 types of optimized built-in algorithm components.
Provides core capabilities such as multiple modes, deep integration with big data engines, multi-framework compatibility, and custom images.
Provides cloud-native AI development, training, and deployment services.

Multiple open-source frameworks

Supports Flink, a stream computing framework.
Supports TensorFlow, PyTorch, Megatron and DeepSpeed, which are optimized deep learning frameworks based on the related open-source frameworks.
Supports Parameter Server, a computing framework that can process hundreds of billions of samples in parallel.
Supports Spark, PySpark, MapReduce, and other mainstream open-source computing frameworks.

Industry-leading AI optimization

Supports high-performance training framework, sparse training scenarios, billions to tens of billions of sparse features, tens to hundreds of billions of samples, and distributed incremental training of thousands of workers.
Supports acceleration of mainstream framework models such as RestNet50 and Transformer+LM by using PAI Blade.

Diverse service modes

Supports fully managed and semi-managed services for public cloud.
Provides high-performance AI computing clusters and lightweight service modes.
Supports periodical scheduling by using DataWorks. You can run scheduled tasks in the production or development environment. This enables data isolation.

Billing

Billing method	Description	Module
Pay-as-you-go	Charged based on the actual usage of each module. Suitable for short-term or uncertain workloads. You pay for resources based on the actual amount of used resources, for example for test environments, development environments, unexpected requirements, or projects in the early phases.	Designer, DSW, DLC, and EAS
Subscription	Subscription for months or years. Suitable for long-term and stable workloads. You pay in advance to use resources for a specific period of time, such as a month or a year. More cost-effective than pay-as-you-go method for long-term use.	DSW and EAS
Resource plan	Quota plans of specific resources that you can purchase in advance. Suitable for scenarios in which you want to use many specific resources. Purchase quota plans at more favorable prices.	DSW
Savings plan	Specific discounts or benefits that you can purchase in advance. Savings plans provide discounted pay-as-you-go rates in exchange for your committing to a specific spending amount within a specific period of time.	DSW and EAS
Pay-by-inference-duration	Charged based on the actual inference duration. The resource specifications support automatic scaling based on the number of service requests. Suitable for inference tasks that require indefinite quantities and is appropriate for high-concurrent requests and dynamic loads.	EAS

Scenarios

Scenario	Use case
LLM applications	Quickly deploy LLMs in EAS One-click deployment of DeepSeek-V3 and DeepSeek-R1
Retrieval-augmented generation (RAG)-based LLM chatbot	Deploy a RAG-based LLM chatbot
AI painting	Use Stable Diffusion web UI to deploy an AI painting service
AI video generation	Generate a video Use ComfyUI to deploy an AI video generation model service
Distributed training	Submit a standalone training job that uses PyTorch

View more scenarios

References

If you are using PAI for the first time, you must activate PAI and create a default workspace. Then, you can refer to the beginers' guide to get started.