AI Potluck
Infrastructure / Core ML frameworks & libraries

Optimum

Hugging Face

Optimum is an extension of Transformers, Diffusers, TIMM, and Sentence-Transformers that provides tools to optimize training and inference on targeted hardware. It supports accelerators and runtimes including ONNX Runtime, OpenVINO, Intel Gaudi, AWS Trainium/Inferentia, and NVIDIA TensorRT-LLM, and handles model export and quantization. It is maintained by Hugging Face.

Verified 2026-08-13 via GitHub and the LICENSE body.

Openness

5 high confidence
5.0
license
Apache-2.0(OSI)
source
public
core-gated
ungated

LICENSE body is verbatim Apache License 2.0; full source public. The repository is public and unarchived, and the README describes the whole library with no paid, enterprise or hosted tier beside it, so source is public and the core ungated.

Adoption

4 high confidence
4.0

1,919,623 PyPI downloads of `optimum` in the trailing 30 days, which falls in the 1M-10M a month band of the software usage scale, level 4.

Capability

4 high confidence
4.0

Broad multi-backend optimization layer, though dependent on the core Transformers stack.

  • https://github.com/huggingface/optimum recorded 2026-08-13

    README still describes hardware-targeted optimization, export and quantization across ONNX Runtime, OpenVINO, Gaudi, Trainium/Inferentia and TensorRT-LLM

Verified 2026-08-13