Apache MXNet

aid: apache-mxnet
name: Apache MXNet
description: Apache MXNet is a retired deep learning framework (now in the Apache Attic) designed for both efficiency and
  flexibility. It provided a multi-language API for building and training deep neural networks with support for distributed
  training, the Gluon high-level API, and deployment on edge devices. MXNet supported Python, Scala, Java, C++, R, Julia,
  and Perl.
type: Index
position: Consumer
access: 3rd-Party
image: https://kinlane-images.s3.amazonaws.com/shared/apis-json/apis-json-logo.jpg
tags:
- AI
- Deep Learning
- Machine Learning
- Neural Networks
- Python
- Retired
created: '2026-03-16'
modified: '2026-04-19'
url: https://raw.githubusercontent.com/api-evangelist/apache-mxnet/refs/heads/main/apis.yml
specificationVersion: '0.19'
apis:
- aid: apache-mxnet:apache-mxnet
  name: Apache MXNet
  description: MXNet provides APIs in Python, Scala, Java, C++, R, Julia, and Perl for deep learning model development, with
    the Gluon high-level API for imperative model building, Symbol/NDArray low-level APIs for efficient computation graphs,
    and distributed training via Parameter Server and Horovod. Final version is 1.9.1.
  humanURL: https://mxnet.apache.org/versions/1.9.1/api
  tags:
  - Deep Learning
  - Distributed Training
  - Gluon
  - Python
  properties:
  - type: Documentation
    url: https://mxnet.apache.org/versions/1.9.1/api
  - type: GettingStarted
    url: https://mxnet.apache.org/versions/1.9.1/get_started
  - type: SDKs
    url: https://pypi.org/project/mxnet/
    title: Python SDK (PyPI)
  - type: SDKs
    url: https://central.sonatype.com/artifact/org.apache.mxnet/mxnet-full_2.12
    title: Scala/Java SDK (Maven)
  - type: GitHubRepository
    url: https://github.com/apache/mxnet
common:
- type: LinkedIn
  url: https://www.linkedin.com/company/apache-mxnet
- type: Portal
  url: https://mxnet.apache.org/
- type: GitHubOrganization
  url: https://github.com/apache
- type: GitHubRepository
  url: https://github.com/apache/mxnet
- type: Wiki
  url: https://cwiki.apache.org/confluence/display/MXNET/Apache+MXNet+Home
- type: IssueTracker
  url: https://issues.apache.org/jira/projects/MXNET/issues
- type: MailingList
  url: mailto:dev@mxnet.apache.org
- type: TermsOfService
  url: https://www.apache.org/licenses/LICENSE-2.0
- type: Features
  data:
  - name: Hybrid Front-End
    description: Seamlessly transitions between Gluon eager imperative mode and symbolic execution for research flexibility
      and production efficiency.
  - name: Distributed Training
    description: Supports Parameter Server and Horovod for scalable distributed training across multiple GPUs and nodes.
  - name: Multi-Language Bindings
    description: Native APIs in Python, Scala, Java, C++, R, Julia, Clojure, and Perl for broad developer accessibility.
  - name: Gluon High-Level API
    description: Intuitive Gluon API for imperative model building with automatic differentiation and dynamic computation
      graphs.
  - name: NDArray API
    description: NumPy-like array operations for GPU-accelerated numerical computing as the foundation of MXNet computations.
  - name: Symbol API
    description: Symbolic computation graph API for efficient inference and production deployment.
  - name: Model Zoo
    description: Pre-trained models for computer vision, NLP, and other tasks accessible via the Gluon model zoo.
  - name: Edge Deployment
    description: Lightweight deployment support for edge devices and mobile platforms via TVM and ONNX export.
- type: UseCases
  data:
  - name: Computer Vision
    description: Build and train image classification, object detection, and segmentation models using GluonCV toolkit.
  - name: Natural Language Processing
    description: Develop NLP models for text classification, sentiment analysis, and language modeling using GluonNLP.
  - name: Time Series Forecasting
    description: Build time series forecasting models using the GluonTS toolkit for probabilistic forecasting.
  - name: Distributed Deep Learning
    description: Train large neural networks across multiple GPUs and nodes using Parameter Server or Horovod.
  - name: Research Prototyping
    description: Rapid prototyping of novel deep learning architectures using the Gluon imperative API.
- type: Integrations
  data:
  - name: GluonCV
    description: Computer vision toolkit built on MXNet providing pre-trained models and training utilities for vision tasks.
  - name: GluonNLP
    description: NLP toolkit built on MXNet with pre-trained language models and text processing utilities.
  - name: GluonTS
    description: Time series modeling toolkit built on MXNet for probabilistic forecasting.
  - name: ONNX
    description: ONNX model format support for importing and exporting models to/from other frameworks.
  - name: TVM
    description: Apache TVM deep learning compiler for optimizing MXNet model deployment on diverse hardware targets.
  - name: Horovod
    description: Horovod distributed training framework integration for efficient multi-GPU and multi-node training.
  - name: D2L.ai
    description: Dive into Deep Learning interactive textbook using MXNet for teaching deep learning concepts.
maintainers:
- FN: Kin Lane
  email: info@apievangelist.com