How to Install PyMOL for Free (Open-Source PyMOL on Windows, Linux and Mac): A Step-by-Step Guide for Students

PyMOL is free if you install the open-source build rather than the paid Incentive build. The fastest route on Windows, Linux and Mac is one conda-forge command: conda install -c conda-forge pymol-open-source. It is published under a BSD-like licence, so unlike the educational build it carries no restriction on research or publication use.

Most students meet PyMOL through a paywall. You search for a download, land on a page asking for a licence file or a subscription, and conclude that the standard tool for looking at protein structures costs money. It does not. Two entirely different products share the name, and the free one is the same rendering engine that produced a large share of the structure figures in the literature. This guide covers the three legitimate free routes, what each one is allowed to be used for, the exact commands per operating system, and a verification test that proves the install actually works before you depend on it in a project. Written by the StemSkills Lab team, from 10+ years in sequence and structural bioinformatics, drug discovery and design, and multiscale molecular modeling.

Is PyMOL actually free, or do you have to pay?

PyMOL is free. Schrödinger maintains a commercial build called Incentive PyMOL that funds development and comes with technical support, and it also publishes the full source code. The source repository at github.com/schrodinger/pymol-open-source states that the software is “published under a BSD-like license”. That licence is what makes the free build usable for a dissertation, a paper figure, or a thesis defence without asking anyone’s permission.

This matters because there is a third option that looks free but is not equivalent. The PyMOL educational licence is offered at no cost to “teachers and high school and college students”, and its terms explicitly exclude academic research and publication. It is built for classroom instruction and homework. If you render a figure with the educational build and put it in a manuscript, you are outside the licence. The open-source build has no such limit, which is why it is the recommended route for anyone doing real work.

Which free PyMOL should you install?

Three routes are genuinely free. They differ in effort, in what you are permitted to do with the output, and in whether you get an installer or a package manager.

The recommendation for a student starting out is conda-forge. It is the only route that behaves identically on all three operating systems, it keeps PyMOL isolated from your system Python, and it puts PyMOL in the same environment as the rest of a computational biology toolchain.

How do you install open-source PyMOL with conda on any operating system?

The conda-forge package pymol-open-source is currently at version 3.1.0 and is built for linux-64, linux-aarch64, linux-ppc64le, osx-64, osx-arm64 and win-64. One recipe therefore covers Windows, Intel Macs, Apple Silicon Macs and Linux.

Step 1: install a conda distribution

If you do not already have conda, install Miniforge. It is the community installer maintained by the conda-forge project itself, and it is configured to pull from conda-forge by default, which is exactly the channel PyMOL lives on. Download the installer for your platform, run it, then close and reopen your terminal so that conda is on your PATH.

Step 2: create a dedicated environment

Do not install PyMOL into your base environment. A separate environment means a dependency conflict in a later project cannot break your visualisation setup.

conda create -n pymol -c conda-forge python=3.11
conda activate pymol

Step 3: install PyMOL

conda install -c conda-forge pymol-open-source

The equivalent channel-pinned form, conda install conda-forge::pymol-open-source, does the same thing and is what the package page lists. Solving the environment takes a few minutes the first time because PyMOL pulls in a graphics stack.

Step 4: launch it

pymol

The graphical interface opens and the terminal prints a startup banner with the version. On every subsequent session you need conda activate pymol first, otherwise the pymol command will not be found.

How do you install PyMOL on Windows without conda?

Two Windows-specific routes exist for people who would rather not manage a Python environment. Both are community-maintained rather than published by Schrödinger, and both are listed on the PyMOL Wiki Windows install page.

The one-click setup. The pymol-open-source-windows-setup project packages open-source PyMOL as a standalone Windows installer that needs no prior Python or conda setup. A portable ZIP build is offered alongside it, and that ZIP bundles the Microsoft Visual C++ Redistributable for Visual Studio 2022 so the binary has what it needs to start. The application installs under C:\Users\<Username>\AppData\Roaming\PyMOL-Open-Source\bin.

Prebuilt wheels. Christoph Gohlke publishes unofficial Windows wheels at cgohlke/pymol-open-source-wheels, covering Python 3.11 through 3.14 on win32, win-amd64 and win-arm64. You install the PyMOL wheel and the separate pymol_launcher wheel with pip, in a command shaped like this:

py.exe -3.14 -m pip install pymol-3.2.0a0-cp314-cp314-win_amd64.whl

Substitute the filename you actually downloaded, because the version and the Python tag must match your interpreter. That repository documents two constraints worth knowing before you start: the Microsoft Visual C++ Redistributable for Visual Studio 2017 to 2026 must be present, and the pymol-open-source package conflicts with the chempy package, so only one of the two can be imported in a single process.

If you are setting up a full Windows docking workstation, the same environment discipline applies to your other tools. Our walkthrough on installing AutoDock Vina on Windows covers the docking half of that setup.

How do you install PyMOL on Linux with a package manager?

Debian and Ubuntu ship a PyMOL package in the standard repositories:

sudo apt update
sudo apt install pymol

On Fedora and its derivatives:

sudo dnf install pymol

This is the least effort of any route, and it is a reasonable choice on a shared lab machine where you do not want to introduce conda. The trade-off is version lag. Distribution packages are frozen at whatever release was current when the distribution version was cut, so an apt install can be a full major version behind the 3.1.0 currently on conda-forge. If a tutorial you are following uses a newer command or a newer plugin, that gap is where the confusion comes from. Check the version in the startup banner before assuming a feature is missing.

How do you install PyMOL on a Mac?

Use the conda-forge route above. It is the cleanest option on macOS because conda-forge builds for both osx-64 and osx-arm64, so the same command gives you a native binary on an Apple Silicon machine rather than an emulated one. Install Miniforge, create the environment, install pymol-open-source, and run pymol.

If PyMOL opens and then quits immediately on macOS, launch it from Terminal rather than from Finder so you can read the error text. A window manager or graphics permission problem produces a silent failure when launched from the desktop and a readable message when launched from a shell.

How do you verify the install actually worked?

An installer finishing without an error is not proof that PyMOL works. The install is only confirmed once it has fetched a structure over the network, rendered it, and written an image file to disk. That exercises the three subsystems that fail independently: the network layer, the OpenGL layer, and the file writer.

Launch PyMOL, then type these into the PyMOL command line one at a time. Ubiquitin, PDB entry 1UBQ, is a good test case because it is small and downloads in a second.

fetch 1ubq
hide everything
show cartoon
spectrum count, rainbow
png ~/pymol_test.png, dpi=300, ray=1

You should see a rainbow-coloured cartoon of a compact protein fold, and pymol_test.png should appear in your home directory. If all five commands run clean, the install is good.

The fetch command pulls directly from the RCSB Protein Data Bank, which as of 26 August 2026 holds 258,735 experimentally determined structures alongside more than a million computed structure models. That archive is the reason a working fetch matters more than a working installer: it is the difference between a visualiser and a visualiser wired into the entire structural biology literature.

To confirm which version conda gave you, run this in the same environment:

conda list pymol-open-source

What can open-source PyMOL not do?

Three honest limitations, so you are not surprised later.

• No vendor technical support. Schrödinger provides support to Incentive subscribers through the subscriber user area. With the open-source build, your support channels are the PyMOL Wiki and community mailing lists. In practice the wiki answers most student questions, but the response is not guaranteed and not immediate.
• No official binary installer from the vendor. Schrödinger publishes source code, not compiled open-source builds. Every convenient binary route listed above comes from conda-forge, a Linux distribution, or a community packager. This is why a plain download link is so hard to find, and why the paywall impression exists in the first place.
• Building from source is a real build. If you compile it yourself, the repository’s INSTALL file requires a C++17 compiler such as gcc 8 or newer, CMake 3.13 or newer, Python 3.9 or newer, and pip with the build module, after which pip install . does the work. Use a prebuilt package unless you specifically need a custom build.

None of these touch rendering quality. The ray tracer, the cartoon representations, the selection algebra and the Python API are the same engine either way.

Which common install errors will you hit, and how do you fix them?

These are the failures that actually stop students, with the fix for each.

• conda: command not found after installing Miniforge. The installer edits your shell startup file, and an already-open terminal has not read it. Close the terminal, open a new one, and try again. On Windows, use the Miniforge Prompt rather than the standard Command Prompt.
• PackagesNotFoundError: pymol-open-source. The conda-forge channel is not being searched. Add it explicitly with -c conda-forge, as in the command above, or use the conda-forge::pymol-open-source form.
• “Solving environment” runs for a very long time or ends in a conflict. You are almost certainly installing into a crowded base environment. Create the fresh environment shown in step 2 and install there. A clean environment with one pinned Python version resolves in a fraction of the time.
• pymol is not found in a new terminal session. The environment is not active. Run conda activate pymol first. This catches people every single session until it becomes reflex.
• PyMOL starts on Windows and closes instantly, or reports a missing DLL. The Microsoft Visual C++ Redistributable is missing. Both Windows wheel routes document this requirement. Install the redistributable, then relaunch.
• PyMOL runs but the graphical window never appears. The Qt interface layer is missing. The PyMOL Wiki documents installing it separately with python -m pip install pyqt5 in the same environment.
• fetch 1ubq fails or hangs. This is a network problem, not a PyMOL problem, and it is common behind a university proxy. Download the PDB file from the RCSB website in a browser and load it from disk instead with load 1ubq.pdb. Everything downstream works identically.

What should you do next, once PyMOL runs?

An installed viewer is a starting point, not a skill. The three things worth learning immediately are selection syntax, which is what makes PyMOL faster than clicking; structural editing, covered in our guide to mutating a residue in PyMOL; and figure preparation with the ray tracer for publication-quality output.

It is also worth installing a second viewer for comparison, because the two tools are good at different things. Our ChimeraX tutorial for beginners covers the main alternative, which handles cryo-EM density maps and very large assemblies more comfortably than PyMOL does.

For where visualisation sits in the wider skill set, and what to learn before and after it, see our computational biology skills roadmap.

Frequently asked questions

Is open-source PyMOL the same software as the paid version?

It is the same core program built from the same publicly released source code. The paid Incentive subscription adds vendor-supplied installers, technical support and subscriber resources. The rendering engine, the Python API and the file format support you use day to day are common to both.

Can I use free PyMOL figures in a published paper or thesis?

Yes, if you use the open-source build, which is released under a BSD-like licence with no restriction on research or publication. You cannot do this with the educational-licence build, whose terms exclude academic research and publication use. Cite PyMOL in your methods section either way.

Do I need to know Python to install or use PyMOL?

No. PyMOL has its own command line and a full graphical interface, and the install is a single conda command. Python becomes useful later, when you want to script repetitive work such as rendering the same view for forty structures, because PyMOL exposes a complete Python API.

Which PyMOL version should a student install in 2026?

Install the current conda-forge release, which is 3.1.0. It matches what recent tutorials assume. Avoid pinning an old version unless a specific plugin you need requires it, and be aware that a Linux distribution package may be considerably older than the conda-forge build.

Does PyMOL run on Apple Silicon Macs?

Yes. conda-forge builds pymol-open-source for osx-arm64 as well as osx-64, so an M-series Mac gets a native build from the standard install command with no emulation layer and no extra configuration.

Is there a completely browser-based option if I cannot install anything?

PyMOL itself is desktop software and has no official web version. If your machine is locked down, the practical answer is to run the conda install inside a cloud notebook environment where you control the kernel, or to use a web viewer for quick inspection and PyMOL for figure work once you have a machine you can install on.