Installation with pip

Note

PyAutoLens requires Python 3.12 or later. On Python 3.9, 3.10 or 3.11, pip install autolens stops with an error telling you to upgrade — it will not quietly install an older release instead. Upgrade Python to 3.12+ before installing.

JAX & GPU

PyAutoLens runs significantly faster on GPUs — often 50x or more compared to CPUs.

This acceleration is achieved through [JAX](https://docs.jax.dev/en/latest/notebooks/thinking_in_jax.html), which provides GPU and TPU support.

JAX is installed by default — a plain pip install autolens includes it (the older pip install autolens[jax] command still works and installs the same thing).

The default install includes CPU-only JAX. To ensure GPU acceleration, it is recommended that you install JAX with GPU support before installing PyAutoLens, by following the official [JAX installation guide](https://jax.readthedocs.io/en/latest/installation.html).

If you install PyAutoLens without a proper GPU setup, a warning will be displayed.

Note

Intel Macs: JAX no longer publishes wheels for Intel (x86_64) macOS, so on these machines pip install autolens automatically installs without JAX and runs on the slower NumPy path — a warning is printed at import to make this clear. Every other supported platform (Windows, Linux, Apple-silicon Macs) gets JAX by default.

Install

We strongly recommend that you install PyAutoLens in a Python virtual environment, with the link attached describing what a virtual environment is and how to create one.

We upgrade pip to ensure certain libraries install:

pip install --upgrade pip

The latest version of PyAutoLens is installed via pip as follows:

pip install autolens

This includes JAX by default, enabling the acceleration described above. If you need an install without JAX on a platform where JAX wheels exist (e.g. a restricted environment), install normally and then remove it:

pip uninstall jax jaxlib

PyAutoLens detects the absence at import and falls back to the fully supported (but much slower) NumPy path.

If pip prints warnings about dependency version conflicts, these can usually be ignored — the instructions below will identify clearly if the installation is a success.

If there are no errors PyAutoLens is installed!

If there is an error check out the troubleshooting section.

Workspace

Next, clone the autolens workspace (the line --depth 1 clones only the most recent branch on the autolens_workspace, reducing the download size):

cd /path/on/your/computer/you/want/to/put/the/autolens_workspace
git clone https://github.com/PyAutoLabs/autolens_workspace --depth 1
cd autolens_workspace

Run the welcome.py script to get started!

python3 welcome.py

It should be clear that PyAutoLens runs without issue.

If there is an error check out the troubleshooting section.

Numba

Numba (https://numba.pydata.org) is an optional library which makes PyAutoLens run a lot faster, which we strongly recommend users have installed.

You can install numba via the following command:

pip install numba

Some users have experienced difficulties installing numba, which is why it is an optional library. If your installation is not successful, you can use PyAutoLens without it installed for now, to familiarize yourself with the software and determine if it is the right software for you.

If you decide that PyAutoLens is the right software, then I recommend you commit the time to getting a successful numba install working, with more information provided at this readthedocs page

Optional

For interferometer analysis there is one optional dependency that must be installed via the command:

pip install nufftax

PyAutoLens will run without these libraries and it is recommended that you only install them if you intend to do interferometer analysis.

If you run interferometer code a message explaining that you need to install these libraries will be printed, therefore it is safe not to install them initially.

Legacy Python versions

We dropped support for Python 3.9, 3.10 and 3.11 in release 2026.7.29.2 (July 2026) — the first release published declaring Requires-Python >=3.12.

Raising that floor does not retract what is already published. Releases at or below 2026.7.29.1 were published declaring >=3.9, and PyPI metadata is immutable, so they remain valid candidates forever. Left alone, pip install autolens on an older Python did not fail — it walked back to 2026.7.29.1 and installed a months-old stack without JAX, reporting nothing.

Release 2026.7.29.1.post1 exists to stop that. It contains no code, declares Requires-Python <3.12, and raises an error when pip tries to build it, so an unsupported Python gets an explanation instead of a stale install.

If you need a historical release, pin it exactly — that still resolves on older Pythons:

pip install autolens==2025.10.6.1

One gap remains: pip install --only-binary=:all: autolens skips source distributions entirely, so it steps past 2026.7.29.1.post1 and installs the old wheel silently. If you use that flag, pin the version you want.