OnnxSlim can help you slim your onnx model, with less operators, but same accuracy, better inference speed.

• 🚀 2025/11/29: Top 1% on PyPI
• 🚀 2025/11/27: OnnxSlim is merged into NVIDIA TensorRT-Model-Optimizer 🤗🤗🤗
• 🚀 2025/05/17: OnnxSlim is merged into HuggingFace optimum 🤗🤗🤗
• 🚀 2025/04/30: Rank 1st in the AICAS 2025 LLM inference optimization challenge
• 🚀 2025/01/28: Achieved 1M downloads
• 🚀 2024/06/23: OnnxSlim is merged into transformers.js 🤗🤗🤗
• 🚀 2024/06/02: OnnxSlim is merged into ultralytics ❤️❤️❤️
• 🚀 2024/04/30: Rank 1st in the AICAS 2024 LLM inference optimization challenge held by Arm and T-head
• 🚀 2024/01/25: OnnxSlim is merged to mnn-llm, performance increased by 5%

Benchmark

Installation

Using Prebuilt

pip install onnxslim

Install From Source

pip install git+https://github.com/inisis/OnnxSlim@main

Install From Local

git clone https://github.com/inisis/OnnxSlim && cd OnnxSlim/
pip install .

How to use

Bash

onnxslim your_onnx_model slimmed_onnx_model

Inscript

import onnx
import onnxslim

model = onnx.load("model.onnx")
slimmed_model = onnxslim.slim(model)
onnx.save(slimmed_model, "slimmed_model.onnx")

For more usage, see onnxslim -h or refer to our examples

Projects using OnnxSlim

NVIDIA/TensorRT-Model-Optimizer	alibaba/MNN
ultralytics/ultralytics	Mozilla/smart_autofill
alibaba/MNN-LLM	huggingface/transformers.js
huggingface/optimum	PaddlePaddle/PaddleOCR
ModelScope/FunASR	CVCUDA/CV-CUDA
THU-MIG/yolov10	sunsmarterjie/yolov12
nndeploy/nndeploy	deepghs/imgutils

References

• onnx-graphsurgeon
• Polygraphy
• onnx-simplifier
• tabulate
• onnxruntime

Contact

Discord: https://discord.gg/nRw2Fd3VUS QQ Group: 873569894