generated: '2026-07-20'
method: searched
source: https://pypi.org/user/neuralmagic/
note: >-
  Neural Magic was acquired by Red Hat (completed January 2025). Its community
  Python libraries were officially deprecated on 2026-06-02 and no longer receive
  updates or support. They remain published on PyPI and are captured here as the
  provider's first-party package surface. Neural Magic's ongoing engineering
  effort is now upstream in the vLLM project under Red Hat.
packages:
  - language: python
    registry: pypi
    name: deepsparse
    version: 1.9.0
    url: https://pypi.org/project/deepsparse/
    source: https://github.com/neuralmagic/deepsparse
    install: pip install deepsparse
    official: true
    status: deprecated
    summary: Sparsity-aware deep learning inference runtime offering GPU-class performance on CPUs, with APIs to integrate ML into applications.
  - language: python
    registry: pypi
    name: sparseml
    version: 1.9.0
    url: https://pypi.org/project/sparseml/
    source: https://github.com/neuralmagic/sparseml
    install: pip install sparseml
    official: true
    status: deprecated
    summary: Libraries for applying sparsification recipes to neural networks to enable faster and smaller models.
  - language: python
    registry: pypi
    name: sparsezoo
    version: 1.9.0
    url: https://pypi.org/project/sparsezoo/
    source: https://github.com/neuralmagic/sparsezoo
    install: pip install sparsezoo
    official: true
    status: deprecated
    summary: Neural network model repository for highly sparse and sparse-quantized models with matching sparsification recipes.
  - language: python
    registry: pypi
    name: sparsify
    version: 1.7.0
    url: https://pypi.org/project/sparsify/
    source: https://github.com/neuralmagic/sparsify
    install: pip install sparsify
    official: true
    status: deprecated
    summary: UI for automatically sparsifying neural networks and creating sparsification recipes for better inference performance and a smaller footprint.