psutil is a cross-platform Python library for retrieving information about running processes and system utilization — CPU, memory, disks, network and sensors — for developers and operators who need that data programmatically from inside Python code rather than from shell output.

What it is

psutil is a BSD-3-Clause licensed Python library, published on PyPI and documented at https://psutil.io, that exposes a single portable API for process and system monitoring. It supports Linux, Windows, macOS, FreeBSD, OpenBSD, NetBSD, Sun Solaris and AIX, and it returns structured objects — svmem, sswap, sdiskusage, sdiskpart, netio, sconn, pconn, pmem, shwtemp, sbattery, scpufreq — instead of text.

The concrete problem it solves is the portability tax of shelling out to platform-specific tools. psutil implements much of the functionality offered by UNIX command line utilities such as ps, top, free, iotop, netstat, ifconfig and lsof, replacing them with Python calls that behave the same way across platforms and do not require parsing locale-dependent output. It is intended mainly for system monitoring, profiling, limiting process resources and managing running processes.

Key capabilities

• CPU inspection through psutil.cpu_percent(interval=1, percpu=True), psutil.cpu_count(logical=False) and psutil.cpu_freq(), which returns an scpufreq object with current, min and max values.
• Memory reporting through psutil.virtual_memory() and psutil.swap_memory(), including swap-in and swap-out counters.
• Disk inspection through psutil.disk_partitions() and psutil.disk_usage('/'), with device, mountpoint, fstype and mount options.
• Network metrics through psutil.net_io_counters(pernic=True) for per-interface byte counts and psutil.net_connections(kind='tcp') for socket state, local and remote addresses, and owning PID.
• Hardware sensors through psutil.sensors_temperatures() and psutil.sensors_battery().
• Per-process introspection through psutil.Process(7055) with name(), exe(), cpu_percent(), memory_info(), net_connections(kind='tcp') and open_files().
• Process enumeration through psutil.process_iter(['pid', 'name']), plus resource limiting and process management as stated use cases.

Who uses it and how

• Large-scale Python deployments: psutil is among the top 100 most-downloaded packages on PyPI, at 390+ million downloads per month.
• Widely embedded as a dependency: 780,000+ GitHub repositories use it, according to the project.
• Monitoring and profiling workflows, where code samples call psutil.cpu_percent(percpu=True) and psutil.Process(...).memory_info() on a running interpreter.
• Process supervision workflows, using process_iter, per-process cpu_percent() and open_files() to inspect what is running.
• Cross-platform teams that must ship the same instrumentation code to Linux, Windows, macOS, BSD, Solaris and AIX hosts.

Getting started

Install with pip install psutil; the README directs readers to platform-specific installation notes and the API overview and API reference at https://psutil.io.

How it compares

This registry entry names no comparable Python package and no paid product, so psutil stands alone here. Its functional overlap is with the UNIX command line tools it reimplements — ps, top, free, iotop, netstat, ifconfig and lsof — which it replaces with a portable library API rather than a command invocation.

When to use it — and when not to

psutil returns measurements; it does not store them. There is no bundled database, no retention policy, no dashboard and no alert routing, so a self-hoster who wants a monitoring stack must write and operate the collection loop, the storage layer and the alerting around it. Teams wanting a turnkey observability platform with a UI should look elsewhere, and anyone who needs a stable long-term API contract should check the release history themselves, since the material provided here gives no version numbers or release cadence, and the project carries 258 open issues.

pip install psutil

>>> import psutil
>>> psutil.cpu_percent(interval=1, percpu=True)
[4.0, 6.9, 3.7, 9.2]
>>> psutil.cpu_count(logical=False)
2
>>> psutil.cpu_freq()
scpufreq(current=931.42, min=800.0, max=3500.0)

>>> psutil.virtual_memory()
svmem(total=10367352832, available=6472179712, percent=37.6, used=8186245120, free=2181107712, ...)
>>> psutil.swap_memory()
sswap(total=2097147904, used=296128512, free=1801019392, percent=14.1, sin=304193536, sout=677842944)

>>> psutil.disk_partitions()
[sdiskpart(device='/dev/sda1', mountpoint='/', fstype='ext4', opts='rw,nosuid'),
 sdiskpart(device='/dev/sda2', mountpoint='/home', fstype='ext', opts='rw')]
>>> psutil.disk_usage('/')
sdiskusage(total=21378641920, used=4809781248, free=15482871808, percent=22.5)

>>> psutil.net_io_counters(pernic=True)
{'eth0': netio(bytes_sent=485291293, bytes_recv=6004858642, ...),
 'lo': netio(bytes_sent=2838627, bytes_recv=2838627, ...)}
>>> psutil.net_connections(kind='tcp')
[sconn(fd=115, family=2, type=1, laddr=addr(ip='10.0.0.1', port=48776), raddr=addr(ip='93.186.135.91', port=80), status='ESTABLISHED', pid=1254),
 ...]

>>> psutil.sensors_temperatures()
{'coretemp': [shwtemp(label='Physical id 0', current=52.0, high=100.0, critical=100.0),
              shwtemp(label='Core 0', current=45.0, high=100.0, critical=100.0)]}
>>> psutil.sensors_battery()
sbattery(percent=93, secsleft=16628, power_plugged=False)

>>> p = psutil.Process(7055)
>>> p.name()
'python3'
>>> p.exe()
'/usr/bin/python3'
>>> p.cpu_percent(interval=1.0)
12.1
>>> p.memory_info()
pmem(rss=3164160, vms=4410163, shared=897433, text=302694, data=2422374)
>>> p.net_connections(kind='tcp')
[pconn(fd=115, family=2, type=1, laddr=addr(ip='10.0.0.1', port=48776), raddr=addr(ip='93.186.135.91', port=80), status='ESTABLISHED')]
>>> p.open_files()
[popenfile(path='/home/giampaolo/monit.py', fd=3, position=0, mode='r', flags=32768)]
>>>
>>> for p in psutil.process_iter(['pid', 'name']):
...     print(p.pid, p.name())
...
1 systemd
2 kthreadd
3 ksoftirqd/0
...