All Products
Search
Document Center

Platform For AI:Official PAI images

Last Updated:Sep 25, 2025

Platform for AI (PAI) provides a range of prebuilt official images with different frameworks and CUDA versions. Select a suitable image when using Deep Learning Container (DLC), Elastic Algorithm Service (EAS), or Data Science Workshop (DSW) to quickly set up an AI development environment. This topic describes the features of PAI official images and provides a list of core images.

Understand the naming convention

PAI official images follow a specific naming convention that helps you identify key details from the name. An image name typically includes the following parts. We recommend following a similar convention when creating custom images.

Sample name

Name breakdown

Module identifier

tensorflow:2.11-gpu-py39-cu112-ubuntu20.04

  • Framework: TensorFlow 2.11

  • Suitable for: GPU instances

  • Language: Python 3.9

  • CUDA version: 11.2

  • OS: Ubuntu 20.04

Official images are tailored for different PAI services:

  • Images with a -training suffix are for DLC.

  • Images with an -inference suffix are for EAS.

  • Images with a -develop suffix are for DSW.

deeprec-develop:2302-tensorflow1.15-cpu-py36-ubuntu18.04

  • Frameworks: TensorFlow 1.15 and DeepRec 2302

  • Suitable for: CPU instances

  • Language: Python 3.6

  • OS: Ubuntu 18.04

Official image features

PAI provides prebuilt images for multiple machine learning frameworks. This section describes the official PAI images for mainstream frameworks. You can view the full list of official PAI images on the AI Computing Asset Management > Images page in the PAI console.

TensorFlow

Framework version

CUDA version (GPU instance types only)

Operating system

  • TensorFlow2.6

  • TensorFlow2.3

  • TensorFlow2.21

  • TensorFlow2.11

  • TensorFlow 1.15 and TensorFlow 1.15.5

  • TensorFlow1.12

  • CUDA 11.4

  • CUDA 11.3

  • CUDA 11.2

  • CUDA 10.1

  • CUDA 10.0

  • Ubuntu 20.04

  • Ubuntu 18.04

TensorFlow Serving

Framework version

CUDA version (GPU only)

Operating system

  • TensorFlowServing2.11.1

  • TensorFlowServing1.15.0

  • CUDA 11.2

  • CUDA 10.0

  • Ubuntu 20.04

  • Ubuntu 18.04

  • Ubuntu 16.04

Pytorch

Framework version

CUDA version (GPU instance types only))

Operating system

  • Pytorch2.1

  • Pytorch2.0

  • Pytorch1.8

  • Pytorch1.7

  • Pytorch1.12

  • Pytorch1.11

  • Pytorch1.10

  • CUDA 12.1

  • CUDA 11.7

  • CUDA 11.4

  • CUDA 11.3

  • CUDA 10.1

  • Ubuntu 22.04

  • Ubuntu 20.04

  • Ubuntu 18.04

DeepRec

Framework version

CUDA version (GPU instance types only))

Operating system

  • DeepRec2302

  • DeepRec2212

CUDA 11.4

Ubuntu 18.04

XGBoost

Framework version

CUDA version (GPU only)

Operating system

XGBoost1.6.0

Not applicable, supports CPU instance types only

Ubuntu 18.04

Triton Inference Server

Framework version

CUDA version (GPU instance types only))

Operating system

  • TritonServer23.02

  • TritonServer21.09

  • CUDA 12.0

  • CUDA 11.4

Ubuntu 20.04

Images for common scenarios

Lingjun Intelligent Computing Service (Serverless Edition)

Image name

Framework

Instance type

CUDA version

Operating system

Supported region

Programming language and version

deepspeed-training:23.06-gpu-py310-cu121-ubuntu22.04

  • PyTorch 2.1

  • Megatron-LM 23.06

  • DeepSpeed 0.9.5

  • Transformers 4.29.2

  • Nemo 1.19.0

GPU

12.1

ubuntu22.04

China (Ulanqab)

Python3.10

megatron-training:23.06-gpu-py310-cu121-ubuntu22.04

  • PyTorch 2.1

  • Megatron-LM 23.06

  • DeepSpeed 0.9.5

  • Transformers 4.29.2

  • Nemo 1.19.0

GPU

12.1

ubuntu22.04

China (Ulanqab)

Python3.10

nemo-training:23.06-gpu-py310-cu121-ubuntu22.04

  • PyTorch 2.1

  • Megatron-LM 23.06

  • DeepSpeed 0.9.5

  • Transformers 4.29.2

  • Nemo 1.19.0

GPU

12.1

ubuntu22.04

China (Ulanqab)

Python3.10

Artificial Intelligence Generated Content (AIGC)

Image name

Framework

Instance type

CUDA version

Operating system

Region

Programming language and version

stable-diffusion-webui:3.0

StableDiffusionWebUI3.0

GPU

11.7

ubuntu22.04

  • China (Hangzhou)

  • China (Shanghai)

  • China (Beijing)

  • China (Zhangjiakou)

  • China (Ulanqab)

  • China (Shenzhen)

  • China (Heyuan)

  • China (Chengdu)

Python3.10

stable-diffusion-webui:2.2

StableDiffusionWebUI2.2

GPU

11.7

ubuntu22.04

Python3.10

stable-diffusion-webui:1.1

StableDiffusionWebUI1.1

GPU

11.7

ubuntu22.04

Python3.10

stable-diffusion-webui-env:pytorch1.13-gpu-py310-cu117-ubuntu22.04

SD-WebUI-ENV

GPU

11.7

ubuntu22.04

Python3.10

EAS deployment

The following table lists some of the official PAI images that you can use in EAS. To view the list of all images, go to the AI Computing Asset Management > Images page in the PAI console. The image addresses in the following table use the China (Hangzhou) region as an example.

Image name

Framework

Image description

Image address

chat-llm-webui:3.0-blade

  • Blade 0.4.3.dev118+gea4a4a1

  • PyTorch 2.1.0+cu118

  • Transformers 4.36.2

Provides an LLM inference service using the Blade backend. Supports both WebUI and API access.

eas-registry-vpc.cn-hangzhou.cr.aliyuncs.com/pai-eas/chat-llm-webui:3.0-blade

chatbot-langchain:1.0

ChatbotLangChain 1.0

For building Knowledge Base Q&A applications on LangChain.

eas-registry-vpc.cn-hangzhou.cr.aliyuncs.com/pai-eas/chatbot-langchain:1.0

comfyui:0.2-api

ComfyUI 0.2

A ComfyUI-based image that provides asynchronous APIs for Text-to-Image and Image-to-Image tasks.

eas-registry-vpc.cn-hangzhou.cr.aliyuncs.com/pai-eas/comfyui:0.2

comfyui:0.2

ComfyUI 0.2

A ComfyUI-based image for Text-to-Image and Image-to-Image use cases.

eas-registry-vpc.cn-hangzhou.cr.aliyuncs.com/pai-eas/comfyui:0.2

comfyui:0.2-cluster

ComfyUI 0.2

A ComfyUI-based image for Text-to-Image and Image-to-Image use cases.

eas-registry-vpc.cn-hangzhou.cr.aliyuncs.com/pai-eas/comfyui:0.2

kohya_ss:2.2

Kohya 2.2

For deploying Stable Diffusion model fine-tuning applications using Kohya.

eas-registry-vpc.cn-hangzhou.cr.aliyuncs.com/pai-eas/kohya_ss:2.2

modelscope-inference:1.9.1

ModelScope 1.9.1

An EAS image for deploying models from the ModelScope library.

eas-registry-vpc.cn-hangzhou.cr.aliyuncs.com/pai-eas/modelscope-inference:1.9.1

stable-diffusion-webui:4.2-cluster-webui

StableDiffusionWebUI 4.2

A Stable Diffusion WebUI-based image for Text-to-Image and Image-to-Image tasks. Supports concurrent access for multiple users with resource isolation.

eas-registry-vpc.cn-hangzhou.cr.aliyuncs.com/pai-eas/stable-diffusion-webui:4.2

stable-diffusion-webui:4.2-api

StableDiffusionWebUI 4.2

A Stable Diffusion WebUI-based image that provides asynchronous APIs for Text-to-Image and Image-to-Image tasks.

eas-registry-vpc.cn-hangzhou.cr.aliyuncs.com/pai-eas/stable-diffusion-webui:4.2

stable-diffusion-webui:4.2-standard

StableDiffusionWebUI 4.2

A Stable Diffusion WebUI-based image for Text-to-Image and Image-to-Image use cases.

eas-registry-vpc.cn-hangzhou.cr.aliyuncs.com/pai-eas/stable-diffusion-webui:4.2

tensorflow-serving:2.14.1

TensorflowServing 2.14.1

Contains TensorFlow Serving and suitable for inference services based on TensorFlow models. This image supports only CPU instances.

eas-registry-vpc.cn-hangzhou.cr.aliyuncs.com/pai-eas/tensorflow_serving:2.14.1

tensorflow-serving:2.14.1-gpu

TensorflowServing 2.14.1

Deploys TensorFlow models as an inference service in GPU environments, based on the open-source TensorFlow Serving framework.

eas-registry-vpc.cn-hangzhou.cr.aliyuncs.com/pai-eas/tensorflow_serving:2.14.1-gpu

chat-llm-webui:3.0

  • PyTorch 2.0.1

  • Transformers 4.33.3

Provides an LLM inference service using the Hugging Face backend. Supports both WebUI and API access.

eas-registry-vpc.cn-hangzhou.cr.aliyuncs.com/pai-eas/chat-llm-webui:3.0

chat-llm-webui:3.0-vllm

  • PyTorch 2.1.2

  • Transformers 4.36.2

  • VLLM 0.2.7

Provides an LLM inference service using the vLLM backend. Supports both WebUI and API access.

eas-registry-vpc.cn-hangzhou.cr.aliyuncs.com/pai-eas/chat-llm-webui:3.0-vllm

huggingface-inference:1.0-transformers4.33

Transformers 4.33

An EAS image for deploying models from the Hugging Face Transformers library.

eas-registry-vpc.cn-hangzhou.cr.aliyuncs.com/pai-eas/huggingface-inference:transformers-4.33

tritonserver:23.11-py3

TritonServer 23.11

Deploys various models as an inference service, based on the open-source Triton Inference Server.

eas-registry-vpc.cn-hangzhou.cr.aliyuncs.com/pai-eas/tritonserver:23.11-py3