ModelArts Service Selection

ModelArts Studio (MaaS)

MaaS offers an end-to-end toolchain for foundation model production, along with Ascend compute resources.

MaaS integrates leading open-source models, including Llama, Yi, Qwen, and DeepSeek. All models are fully adapted and optimized for AI Cloud Service, resulting in improved accuracy and performance. You no longer need to build models from scratch; instead, you can simply choose suitable pre-trained models for fine-tuning or direct application, reducing the workloads of model integration.

MaaS helps you modify and apply built-in open-source models on a MaaS platform through tasks like fine-tuning, setting up agents, or reconstructing the model layer.

The UI is straightforward. You do not need to write model code. Built-in models are ready for direct use.

There is no need to understand NPU compute; simply access the provided APIs.

ModelArts Standard

This one-stop platform simplifies AI development with a user-friendly UI. It offers tools for data management, development environments, model training, model management, and deployment. Everything you need for seamless AI development, training, and application is included.

ModelArts Standard allows you to independently develop AI. You must prepare data, algorithms, and models. The platform only offers AI development tools and features.

The UI is straightforward, offering comprehensive features with diverse AI capabilities.

Unlike MaaS, this product supports a broader variety of models.

ModelArts Lite Server

ModelArts Lite Server offers cloud servers running on bare metal servers or elastic cloud servers, accessible via EIPs.

ModelArts Lite Server offers affordable AI compute for users with existing AI development platforms. It includes popular AI kits and Huawei's acceleration plug-ins.

BMSs are flexible.

Fault detection and recovery are provided.

ModelArts Lite Cluster

ModelArts Lite Cluster provides direct access to Kubernetes APIs, enabling you to manage nodes and clusters within resource pools.

This product works best for running AI platforms or services on Kubernetes when you only need compute. You should have basic Kubernetes knowledge.

Standard Kubernetes for fast cloud-native migration

ModelArts Studio (MaaS)

NPU

Users are agnostic to it.

OBS

Users are agnostic to it.

Yearly/Monthly

Pay-per-use

Token-based billing

ModelArts Standard

NPU

GPU

Users are agnostic to it.

Public resource pool: OBS

Dedicated resource pool: SFS Turbo and OBS

Users are agnostic to it.

Yearly/Monthly

Pay-per-use

ModelArts Lite Server

NPU

GPU

Huawei Cloud EulerOS

Ubuntu

SFS Turbo, OBS, and EVS

For details, see:

Configuring the Storage

For details, see:

Mapping Between Compute Resources and Image Versions

Yearly/Monthly

ModelArts Lite Cluster

NPU

GPU

EulerOS

Huawei Cloud EulerOS

SFS Turbo, OBS, and EVS

For details, see: Configuring Lite Cluster Storage

For details, see:

Software Versions Required by Different Models

Yearly/Monthly

ModelArts Studio (MaaS)

MaaS console

UI

CN-Hong Kong

ModelArts Standard

ModelArts console

UI

All Huawei Cloud regions

ModelArts Lite Server

ModelArts console

Create a Lite Server node through the UI or API. Then, log in to the node using SSH.

CN-Hong Kong, ME-Riyadh, CN East 2, and CN Southwest-Guiyang1

ModelArts Lite Cluster

ModelArts console

Create a Lite Cluster node through the UI. Then, log in to the node using SSH.

All Huawei Cloud regions