onnxruntime-gpu has 4.2M monthly downloads and 21K GitHub stars. Version 1.27.0 supports Python 3.11+, first published in 2018.
It is actively maintained under a permissive license, last released in Jun 2026.
It is built on top of flatbuffers, numpy and packaging. It is used by 156 tracked packages, including inference-gpu, foundry-local-sdk and setiastrosuitepro.
GitHub topics include deep-learning, onnx and neural-networks.
Package Insights
((week_daily_avg - month_daily_avg) / month_daily_avg) * 100Weekly Downloads
GitHub Stars
Downloads by OS
Python Versions
Top Countries
Dependencies
- flatbuffers
- numpy >=1.21.6
- packaging
- protobuf >=4.25.8
7 optional dependencies
- ml-dtypes[quantization]
- nvidia-cuda-nvrtc-cu13[cuda]
- nvidia-cuda-runtime-cu13[cuda]
- nvidia-cudnn-cu13[cudnn]
- nvidia-cufft-cu13[cuda]
- nvidia-curand-cu13[cuda]
- sympy[symbolic]
Used By
- inference-gpu
- foundry-local-sdk
- setiastrosuitepro
- memolib
- fastembed-gpu
- onnxruntime-genai-cuda
- vhup
- inaspeechsegmenter
- embed-anything-gpu
- oemer
- toposync-vision-cuda
- nucleofind
- yta-tts-kokoro-onnx
- ultimate-rvc
- maai
- catt-tashkeel
- visagene
- svr-tts
- seraph-db-onnx
- animl
- megaparse
- kiri-ocr
- zim-anything
- opentau
- forensicface
- socaity-face2face
- anylabeling-gpu
- bert-onnx
- triton-model-navigator
- yolov9-onnx
- indic-asr-onnx
- ai-models-panguweather-gfs
- torchness
- orient-express
- kelp-o-matic
- ai-models-panguweather
- onnx-paddleocr
- pyethnicity
- nemo2riva
- cleanframes
- habitat-mapper
- easy-dwpose
- yolov10-onnx
- sparrow-engine-gpu
- audiozen
- vogent-turn
- soulxpodcast
- safeai-face
- juxtapose
- densecall