h5py/h5py

Health v2 · maintenance only

Adoption not part of the score

2249 stars · 564 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

h5py is a Pythonic wrapper around the HDF5 binary data format, built on a Cython binding of the HDF5 C API. It lets you store and manipulate huge amounts of numerical data from NumPy, including slicing multi-terabyte datasets on disk as if they were in-memory arrays.

Use cases

• store large numerical datasets in HDF5 files from Python
• read and slice multi-terabyte datasets on disk like NumPy arrays
• exchange scientific data with MATLAB or IDL users
• organize thousands of datasets in a single tagged file
• access low-level HDF5 C API features from Python
• work with HDF5 files with MPI parallel support

When to choose

• you need to persist large NumPy arrays or hierarchical numerical data in a standard binary format
• you want out-of-core access to datasets too big for memory
• you need interoperability with other HDF5 tools like MATLAB, IDL, or PyTables

When to avoid

• you only need simple tabular data where CSV, Parquet, or SQLite suffice
• you need a database with querying and indexing rather than a file format
• you work outside the Python/NumPy ecosystem and can use the HDF5 C library directly

Facets

library · maturity stable

serialization file-system data-science data-science files developer-tools python cross-platform windows hdf5 numpy binary-data-format scientific-computing large-datasets linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="h5py/h5py")