---
title: "Installation"
description: "Install cpbenchy and the solvers you want to compare, and check that your machine is ready."
---

> Documentation Index
> Fetch the complete documentation index at: https://docs.cpbenchy.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Installation

cpbenchy needs Linux and Python 3.10 or later.

1. **Install cpbenchy**

   ```sh
   pip install cpbenchy      # or: uv add cpbenchy
   ```

   It comes with CPMpy and OR-Tools.
2. **Install the solvers you want to compare**

   Install additional solvers through [CPMpy](https://cpmpy.readthedocs.io/en/latest/modeling.html):

   ```sh
   pip install cpmpy[exact,highs]
   ```
3. **Check your machine**

   ```sh
   cpbenchy doctor
   ```

   ```
   cpbenchy 0.1.0, cpmpy 1.1.0, python 3.13.11
   runexec:  ok, runs get cgroup limits and measurements (executor: runlimit)
   machine:  20 CPUs, 14 physical cores, 1 NUMA node(s), 31717 MiB memory
   solvers:  ortools 9.15.6755, exact 2.2.1, highs 1.14.0
             not installed: z3, minizinc, gcs, pysat, ...
   ```

   `runexec: ok` means cpbenchy can measure runs reliably. If it says otherwise, cpbenchy still
   works, with less precise limits: see [Measurement and limits](/guides/limits/#executors).

## Alpha

cpbenchy is in alpha: until version 1.0, a new version may change commands, options, the Python API
or the result format. Pin the version in projects that depend on it, and read the changes before
upgrading:

```sh
pip install "cpbenchy==0.1.*"      # or: uv add "cpbenchy==0.1.*"
```

Results record the cpbenchy version that made them, in `run.json`.

Next: [your first benchmark](/getting-started/first-benchmark/).

Source: https://docs.cpbenchy.com/getting-started/installation/index.mdx