How to install NumPy with pip

NumPy belongs in the same Python environment that runs the project, notebook, or script. pip is the environment-based installer for Python projects that already use Python and do not need Conda or a project manager such as uv.

A local venv directory keeps the interpreter, pip, and the NumPy install together. Running python -m pip after activation ties the installer to the active interpreter instead of whichever pip command appears first on PATH.

A finished setup should show pip coming from .venv, report a NumPy version, and create a small array. Importing NumPy and executing array code catch mismatched interpreters and extension loading failures before notebooks, scripts, or IDEs depend on the environment.

1. Open a terminal in the project directory with Python 3 available.

   If python3 -m venv reports that ensurepip or venv is missing on Debian or Ubuntu, install the distribution's python3-venv package, then rerun the virtual environment command.

2. Create a virtual environment in the project directory.

   $ python3 -m venv .venv

   On Windows, use py -m venv .venv.

3. Activate the virtual environment.

   $ . .venv/bin/activate

   On Windows, run .venv\Scripts\activate from the project directory.

4. Confirm that pip belongs to the active virtual environment.

   $ python -m pip --version
   pip 25.1.1 from /home/user/project/.venv/lib/python3.14/site-packages/pip (python 3.14)

   The path should include .venv. A different path means the shell is not using the project environment.

5. Install NumPy with pip.

   $ python -m pip install numpy
   Collecting numpy
   ##### snipped #####
   Successfully installed numpy-2.5.0

   pip downloads the current compatible wheel from PyPI when one is available. The exact version changes as NumPy releases new packages for the active Python version and platform.

6. Confirm that Python can import NumPy.

   $ python -c "import numpy as np; print(np.__version__)"
   2.5.0

   The version may differ when PyPI has a newer compatible NumPy release. Any traceback means the active interpreter still cannot import the installed package.
   Related: Fix import errors

7. Run a small NumPy array calculation.

   $ python -c "import numpy as np; print(np.arange(3).tolist())"
   [0, 1, 2]

   The array output confirms that NumPy imports and executes basic array code inside the same environment.