PyPI statistics
We all love stats, so here are some useful statistics about PyPI.
First-party statistics
These statistics are provided directly by PyPI.
- PyPI provides public datasets, including download statistics and metadata via BigQuery
- PyPI provides a public dashboard with statistics on usage and performance
Third-party statistics
These statistics are provided by other third-party services unaffiliated with PyPI.
Top projects by total package size
Here is a list of the top 100 projects based on the sum of their packages' sizes (in bytes). This page is cached for 24 hours, so don't expect the numbers to be realtime.
| Project name | Sum of release files (bytes) |
|---|---|
| All of PyPI | 45.3 TB |
| tf-nightly | 681.4 GB |
| tf-nightly-cpu | 531.4 GB |
| tensorflow | 421.1 GB |
| torch | 271.4 GB |
| tf-nightly-cpu-aws | 269.1 GB |
| lalsuite | 257.4 GB |
| tensorflow-gpu | 237.9 GB |
| catboost-dev | 237.0 GB |
| ddtrace | 210.0 GB |
| tensorflow-io-nightly | 205.4 GB |
| frida | 201.7 GB |
| paddlepaddle-gpu | 190.9 GB |
| tensorflow-cpu | 175.2 GB |
| openvisus | 171.5 GB |
| tf-nightly-macos | 139.9 GB |
| litert-converter | 139.0 GB |
| tf-nightly-intel | 116.9 GB |
| torchmonarch | 107.3 GB |
| ray | 104.0 GB |
| grpcio | 93.6 GB |
| nvidia-cudnn-cu12 | 93.1 GB |
| uv | 87.8 GB |
| scipy | 84.7 GB |
| duckdb | 75.0 GB |
| tensorflow-rocm | 74.8 GB |
| cupy-cuda92 | 71.2 GB |
| cupy-cuda100 | 70.2 GB |
| cupy-cuda90 | 66.0 GB |
| grpcio-tools | 66.0 GB |
| ovito | 64.4 GB |
| catboost | 64.1 GB |
| pulumi-azure-native | 60.6 GB |
| jaxlib | 59.4 GB |
| pyAgrum-nightly | 59.0 GB |
| onnxruntime-gpu | 58.7 GB |
| cupy-cuda91 | 55.5 GB |
| cupy-cuda101 | 55.0 GB |
| numpy | 55.0 GB |
| cmake | 52.1 GB |
| mindspore | 51.3 GB |
| semgrep | 51.1 GB |
| homeassistant | 51.0 GB |
| chdb | 49.6 GB |
| vtk | 49.1 GB |
| claude-agent-sdk | 48.8 GB |
| pyarrow | 48.5 GB |
| opencv-contrib-python | 48.1 GB |
| rerun-sdk | 47.1 GB |
| tensorflow-intel | 45.6 GB |
| home-assistant-frontend | 44.1 GB |
| ruff | 43.7 GB |
| Panda3D | 43.6 GB |
| llama-cpp-cffi | 43.5 GB |
| tf-nightly-tpu | 42.9 GB |
| opencv-python | 42.3 GB |
| opencv-contrib-python-headless | 41.8 GB |
| casadi | 41.6 GB |
| botocore | 41.1 GB |
| bpy | 39.3 GB |
| icechunk | 39.3 GB |
| pantsbuild.pants | 38.7 GB |
| vllm | 38.7 GB |
| awscrt | 37.4 GB |
| libtpu | 37.2 GB |
| fbgemm-gpu | 36.6 GB |
| deepspeech-gpu | 36.6 GB |
| cupy-cuda80 | 36.3 GB |
| amalgam-lang | 36.1 GB |
| cupy-cuda102 | 35.5 GB |
| lmdeploy | 35.5 GB |
| rasterio | 35.5 GB |
| aws-cdk-lib | 34.8 GB |
| ray-cpp | 34.7 GB |
| sickrage | 34.2 GB |
| docling-parse | 33.7 GB |
| polars | 33.2 GB |
| zeroconf | 33.2 GB |
| pyqt5-tools | 32.2 GB |
| llvmlite | 32.1 GB |
| opencv-python-headless | 31.6 GB |
| rdkit | 31.5 GB |
| pulsar-client-sn | 31.2 GB |
| nvidia-cudnn-cu13 | 31.1 GB |
| paddlepaddle | 30.7 GB |
| passagemath-gap | 30.7 GB |
| sgl-kernel | 30.5 GB |
| passagemath-macaulay2 | 30.4 GB |
| matplotlib | 30.4 GB |
| passagemath-modules | 30.4 GB |
| PySide6-Addons | 30.2 GB |
| chalkdf | 30.1 GB |
| nautilus-trader | 29.5 GB |
| pypowsybl | 29.5 GB |
| pylance | 29.4 GB |
| spacy | 29.4 GB |
| tensorflow-io-2.0-preview | 29.3 GB |
| passagemath-linbox | 29.2 GB |
| xformers | 29.1 GB |
| openvino | 28.9 GB |
| passagemath-giac | 28.6 GB |
