Guide · Windows 11 and 10

What is a normal CPU, GPU and SSD temperature?

A hot number is not always a problem and a cool one is not always safe. Here are the normal ranges, what throttling is, and how to watch your PC.

Updated 5 October 2026 · 5 min read

Normal temperature ranges

These are typical ranges, not limits. Every chip has its own maximum, which its maker publishes, and laptops run hotter than desktops with the same chip.

• CPU: roughly 30 to 50 °C when idle, 50 to 75 °C in everyday use, and 70 to 90 °C or more under a sustained heavy load such as rendering or a demanding game. Intel says the maximum junction temperature of most mainstream Core processors is 100 °C, though it varies by model and is listed on each chip's ARK page. Many AMD Ryzen 7000 chips are designed to boost until they reach about 95 °C (89 °C for the X3D models), so a reading in the 80s or low 90s under load is expected there.
• GPU: roughly 30 to 50 °C idle and 60 to 85 °C while gaming. The slowdown point is commonly in the 80s to low 90s depending on the card, so check your card's specification.
• NVMe SSD: roughly 30 to 50 °C idle and 50 to 70 °C under sustained transfers. A drive reports its own warning temperature, typically around 70 to 80 °C on consumer drives, and a higher critical one. SATA SSDs usually run cooler.
• Hard drive: as a rule of thumb, keep it below about 50 °C. Check the maker's specification for the exact operating range.

Judge a temperature by its behavior: how fast it climbs, where it settles under your usual load, and whether it has changed since last month. A PC that settles at 82 °C in games and always has is fine. One that used to settle at 70 °C and now reaches 90 °C needs cleaning or a new fan.

What throttling is

When a chip nears its limit it lowers its own clock speed and power to stay under it. This is the protection working, not damage. You feel it as frame rates that fall after a few minutes, a build or export that slows, or fans at full speed. If the temperature keeps rising anyway, the system shuts itself off to protect the hardware. Throttling is only a problem if it is constant, because then you are paying for performance you never receive.

Why Windows often cannot show your CPU temperature

Windows has no built-in, universal CPU temperature display. Task Manager shows CPU use and speed, but not temperature. Real core temperatures are read from sensors inside the processor by a small kernel driver, which is what most hardware monitor programs install. Without one, you are left with what the firmware exposes to Windows: ACPI thermal zones. These are whatever the PC maker chose to publish, often a single reading that is not the CPU core, and some PCs publish none at all.

The graphics driver is a separate source, and Task Manager shows GPU temperature for many cards under Performance, then GPU. Drives report their own temperature through SMART or the NVMe health log.

To see the thermal zones yourself, run this in an administrator PowerShell window. The value is in tenths of a kelvin, so this converts it to degrees Celsius:

Get-CimInstance -Namespace root/wmi -ClassName MSAcpi_ThermalZoneTemperature | Select-Object InstanceName, @{n='Celsius';e={$_.CurrentTemperature/10-273.15}}

If it returns an error or nothing, your firmware does not expose a zone. The UEFI/BIOS screen on most boards shows a CPU temperature too, which is useful as a check at idle.

If it runs too hot

1. Clean it. Dust in the heatsink and filters is the most common cause. Use compressed air with the fans held still, with the PC off.
2. Check airflow. Keep case panels and vents clear. On a laptop, use a hard surface or a stand, not a bed or lap.
3. Find what is busy. A program stuck at high CPU use in the background heats the PC at idle. Start with Task Manager and your startup programs.
4. Check the fans. Confirm every fan spins, and look at the fan curve in your UEFI/BIOS or the maker's utility.
5. Limit the load. Capping the frame rate or switching a laptop to a balanced power mode lowers heat with little visible change.
6. Replace old thermal paste if a PC that has been fine for years starts to run hot. That is a job for confident hands, and is not needed for a new PC.

In Cosmo Cleaner

Release 1.2.0 adds a Temperatures tab. It shows every temperature Windows exposes without a kernel driver, grouped as CPU, GPU, storage, motherboard, battery, memory and other, each marked Normal, Warm, Hot or Critical, using the limits your hardware reports and sensible defaults where it reports none. A sensor that is not exposed is labeled Not exposed, not guessed. Cosmo Cleaner does not install a kernel driver and never loads a third-party one, so per-core CPU temperatures and motherboard sensors appear only when Windows or the PC maker already expose them.

History charts readings over time and keeps the daily minimum, average and maximum per group, so you can see whether the PC runs hotter than it did last month.

Notifications lets you draw a curve. Each point on it means "at or above this temperature for this long, tell me". A hot reading gets a short wait and a warm one a long wait, and at the critical limit Cosmo Cleaner tells you immediately. Start from a preset, override a single sensor, keep alerts quiet while a game runs full screen, and send a test notification to see how it looks. Alerts arrive as Windows notifications, and everything is evaluated on your PC.

Questions