NVIDIA Model Optimizer

license

Apache-2.0(OSI)

source

public

core features withheld

no

The LICENSE file contains the Apache-2.0 text. The repository is public and the README describes no paid tier, enterprise edition or license-gated build, so the core is ungated.

• https://raw.githubusercontent.com/NVIDIA/Model-Optimizer/main/LICENSE recorded 2026-09-30

  LICENSE file is the verbatim Apache-2.0 text, opening with "Apache License" and "Version 2.0, January 2004" and "TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION".

• https://raw.githubusercontent.com/NVIDIA/Model-Optimizer/main/pyproject.toml recorded 2026-09-30

  pyproject.toml declares the license "Apache-2.0" and depends only on plain PyPI packages, such as "nvidia-ml-py>=12" and "torch>=2.8", with the comment "# for faster building of C++ / CUDA extensions". Searched the body for enterprise, license key, proprietary, closed source and paid: none.

• https://raw.githubusercontent.com/NVIDIA/Model-Optimizer/main/README.md recorded 2026-09-30

  README states "NVIDIA Model Optimizer** (referred to as **Model Optimizer**, or **ModelOpt**) is a library comprising state-of-the-art model optimization" techniques and carries a license badge "License-Apache%202.0". Searched the body for enterprise, license key, proprietary, closed source and paid: none.

PyPI downloads of `nvidia-modelopt` are the adoption signal here.

• https://pypistats.org/api/packages/nvidia-modelopt/recent recorded 2026-08-18

  last_month downloads = 418,452 for nvidia-modelopt

NVIDIA Model Optimizer applies six optimization families - quantization, pruning, neural architecture search, distillation, sparsity and speculative decoding - to Hugging Face, PyTorch and ONNX models, exporting checkpoints for TensorRT-LLM, TensorRT, vLLM and SGLang. That breadth of technique sits one step below the retargetable, multi-vendor compilation Apache TVM performs across its whole pipeline.

• https://raw.githubusercontent.com/NVIDIA/Model-Optimizer/main/README.md recorded 2026-09-30

  README says "Model Optimizer currently supports inputs of a [Hugging Face](https://huggingface.co/), [PyTorch](https://github.com/pytorch/pytorch) or [ONNX](https://github.com/onnx/onnx) model." and "Model Optimizer provides Python APIs for users to easily compose the above model optimization techniques and export an optimized quantized checkpoint."

• https://raw.githubusercontent.com/NVIDIA/Model-Optimizer/main/README.md recorded 2026-09-30

  README says the quantized checkpoint "is ready for deployment in downstream inference frameworks like" SGLang, TensorRT-LLM, TensorRT, or vLLM, and lists a techniques table with "Post Training Quantization", "Pruning" and "Distillation".