GPU Temperature Range | Safe, Ideal, & Tolerance Limits 2026
A safe GPU temperature is 30–50°C (86–122°F) at idle and 65–85°C (149–185°F) while gaming. A high GPU temperature starts around 85–90°C (185–194°F) on most NVIDIA RTX 50 cards, while Radeon cards can run up to 110°C (230°F) on the hotspot sensor.
This guide also covers GPU temperature in Fahrenheit, laptop temperature, CPU vs GPU limits and what to do if your GPU is overheated.
Key Takeaways
| State | Safe Range (°C) | Safe Range (°F) | Act When |
|---|---|---|---|
| Idle | 30–50°C | 86–122°F | Stays above 60°C (140°F) |
| Gaming / heavy load | 65–85°C | 149–185°F | Above 85°C (185°F) for long periods |
| NVIDIA throttle zone | 85–90°C, varies by model | 185–194°F | Clocks drop or FPS stutters |
| AMD hotspot limit | Up to 110°C | Up to 230°F | Throttling at fan max |
Table of ContentsShow or hide all 12 sectionsShowHide
How Does GPU Temperature Increase?
Every watt a graphics card uses ends up as heat. Modern cards also boost their clock speed automatically until they hit a power or temperature limit, so they naturally run warm. The main causes of higher temperatures:
- High GPU load: 4K, ray tracing and uncapped frame rates keep the card near 100% usage.
- Poor airflow: If hot air stays inside the case, the GPU keeps breathing its own heat.
- Dust: Clogged fins and fans reduce cooling.
- Warm room: Ambient temperature directly raises GPU temperature.
- Old thermal paste or pads: Heat transfer gets worse after a few years.
- Overclocking: More speed and voltage create more heat.
Safe GPU Temperatures Under Load
Under load, 65–85°C (149–185°F) is safe for most graphics cards, and below 80°C (176°F) is ideal. A high GPU temperature is one that stays above 85°C (185°F) for long periods or close to your model’s maximum.
| Temperature | Status |
|---|---|
| Below 70°C (158°F) | Excellent, plenty of headroom |
| 70–80°C (158–176°F) | Normal for modern games |
| 80–85°C (176–185°F) | Acceptable, watch airflow |
| 85–90°C (185–194°F) | Hot, near the limit for many NVIDIA cards |
| 90°C+ (194°F+) | Too hot, throttling likely |
Maximum GPU Temperature by Brand (NVIDIA, AMD, Intel)
| Card | Limit / Typical Reading | What to Know |
|---|---|---|
| NVIDIA RTX 5090 | 90°C (194°F) | Official maximum GPU temperature |
| NVIDIA RTX 5080 / 5070 Ti | 88°C (190°F) | Official maximum GPU temperature |
| NVIDIA RTX 5070 | 85°C (185°F) | Lowest ceiling of the RTX 50 cards listed here |
| AMD Radeon RX 7000 | Hotspot up to 110°C (230°F) | Junction (hotspot) temperature at this level is within spec, according to AMD |
| AMD Radeon RX 9070 XT | Memory limit about 108°C (226°F) | As read from the GPU by HWiNFO, not an official AMD spec page |
| Intel Arc B580 | About 60–73°C (140–163°F) under gaming load | Review readings, not an official limit. Check Intel’s page for its limit |
These numbers are the point where the card starts to throttle, not the point where it breaks. AMD Radeon cards are judged on the junction (hotspot) sensor, while NVIDIA lists one maximum GPU temperature per model. Always check your exact model’s spec sheet.
Idle temperature matters less than load temperature. A card that idles at 35°C (95°F) but hits 92°C (198°F) in games has a cooling problem. Play a demanding game for 20–30 minutes and note the stable temperature.
Laptop GPU and CPU Temperature
Laptops run hotter than desktops because the cooling system is thin and the CPU and GPU often share the same heat pipes. Typical laptop temperature ranges:
| State | Laptop CPU | Laptop GPU |
|---|---|---|
| Idle / light use | 35–55°C (95–131°F) | 35–55°C (95–131°F) |
| Gaming | 70–85°C (158–185°F), brief peaks to 90°C+ (194°F+) are seen on thin models | 60–85°C (140–185°F) |
If your laptop stays above these ranges, clean the vents, use it on a hard flat surface, try a cooling pad and switch to a lower performance mode. Laptop designs differ a lot, so check the maker’s guidance for your model.
GPU Temperature in Fahrenheit: Quick Conversion Chart
Most monitoring tools show Celsius by default. Use this chart to read your GPU temperature in Fahrenheit. To convert yourself, multiply °C by 1.8 and add 32.
| °C | °F | What It Means |
|---|---|---|
| 30°C | 86°F | Normal idle |
| 50°C | 122°F | Upper end of normal idle |
| 60°C | 140°F | High for idle, check background usage |
| 70°C | 158°F | Excellent under load |
| 80°C | 176°F | Normal under heavy load |
| 85°C | 185°F | Upper end of safe gaming range |
| 90°C | 194°F | Throttle zone for most NVIDIA cards |
| 110°C | 230°F | Radeon hotspot limit |
How to Check GPU Temperature?
Windows Task Manager (Performance tab) shows a basic GPU temperature, but it cannot show your CPU temperature. For more detail on both, use one of these free tools. HWiNFO and MSI Afterburner can show CPU and GPU temperatures side by side, so you can check your whole PC at once.
1. HWiNFO
HWiNFO is the most complete option. Open it in Sensors-only mode, scroll to your GPU section and look for GPU Temperature, GPU Hot Spot and GPU Memory Junction. It also logs readings to a file. Which sensors appear depends on your card and driver: Radeon and older NVIDIA cards usually show hotspot, but NVIDIA RTX 50 cards do not expose hotspot in most standard monitoring tools.
2. GPU-Z
GPU-Z is light and fast. Open the Sensors tab to see temperature, fan speed, clocks, power draw and load. Hotspot and memory readings depend on your card and driver.
3. MSI Afterburner
Afterburner shows live stats inside your game through the RivaTuner overlay. It also controls fan curves and undervolting. Like other tools, it may not show hotspot on RTX 50 cards. Core temperature, fan speed and clock behavior are still reliable signs of a cooling problem.
What is Thermal Throttling?
Thermal throttling is a built-in safety feature in graphics cards. When the GPU gets too hot, it automatically lowers its clock speed, voltage and power draw to bring the temperature back down. This protects the chip and other parts from heat damage, but the cost is lower performance. NVIDIA cards throttle at or just below the maximum temperature listed on their spec sheet, while Radeon cards throttle when the junction (hotspot) temperature nears 110°C (230°F).
Short answer: Thermal throttling is when your GPU slows itself down because it is too hot. You notice it as sudden FPS drops and stuttering.
Signs of thermal throttling:
- FPS that drops after 10–15 minutes of gaming
- Stuttering in heavy scenes
- Clock speeds falling below normal while fans run at full speed
- Black screen or restart if cooling fails completely
How to fix it: Clean dust, improve case airflow, raise the fan curve or undervolt the card.
Throttling can look like a hardware bottleneck, because both show up as low FPS. If your temperatures are fine but FPS is still low, run your parts through a bottleneck calculator to see whether your CPU and GPU are well matched.
GPU Core Temperature vs Hotspot vs Memory Junction
Monitoring software can show three different GPU temperatures, and they measure different things.
- Core (edge) temperature is the overall temperature of the graphics chip. It is the number most tools show first.
- Hotspot (junction) temperature is the hottest single point on the chip, so it is always higher than the core reading. NVIDIA RTX 50 cards do not report it in most standard tools.
- Memory junction temperature is the temperature inside the VRAM chips. It matters most on cards with fast GDDR6X or GDDR7 memory.
Short answer: Core temperature is the main chip reading, hotspot is the hottest point on the chip, and memory junction is the heat inside the VRAM.
| Sensor | What It Shows | Normal Under Load |
|---|---|---|
| Core (edge) | Overall chip temperature | 65–85°C (149–185°F) |
| Hotspot (junction) | Hottest point on the chip | 10–20°C (18–36°F) above core on NVIDIA RTX 40 and older; up to 25–30°C (45–54°F) can be normal on some Radeon cards |
| Memory junction | Temperature inside VRAM | 80–95°C (176–203°F) on GDDR6X / GDDR7 |
What to watch:
- If your card reports hotspot and the gap between core and hotspot stays above 30°C (54°F), the cooler contact, thermal paste or pads may need attention.
- For memory, aim to stay below about 95°C (203°F). GDDR6X chips are built to tolerate 100–110°C (212–230°F) and throttle near 110°C (230°F), but cooler is better for long-term health. The RX 9070 XT has a memory limit of about 108°C (226°F).
- Memory heat rises with room temperature, so a card that is fine in winter can run hot in summer.
- On RTX 50 cards, hotspot is not visible in most tools, so a normal core reading does not guarantee the hottest point is cool. Watch fan speed, clock drops and stability as well.
GPU Undervolting and Power Management
Undervolting means reducing the voltage supplied to your GPU while keeping its clock speed the same or very close. Graphics cards often ship with more voltage than the chip actually needs, so lowering it cuts power draw and heat without a big loss in speed. Power management goes one step further by limiting how many watts the card can use, for example with a lower power limit or a frame rate cap. Together, they make the card run cooler, quieter and more efficient.
Short answer: Undervolting lowers the voltage your GPU uses while keeping similar clock speeds. It usually cuts temperature by 5–15°C (9–27°F) and saves 20–40W with little to no FPS loss.
In one test, lowering an RTX 5090 from 575W to 450W cost only about 5% performance.
How to undervolt safely:
- Open MSI Afterburner and press Ctrl+F for the voltage/frequency curve.
- Pick a lower voltage point at your normal boost clock and flatten the curve after it.
- Apply, then test in a demanding game or benchmark for 20–30 minutes.
- If it crashes, raise the voltage a little and test again.
- Optionally cap your frame rate and lower the power limit by 5–15%.
Results differ from card to card, so treat these numbers as typical ranges, not promises. If you are also planning a GPU upgrade, use a PSU wattage calculator to check your power supply can handle it. NVIDIA lists 1000W of system power for the RTX 5090, for example.
GPU Overheated? What to Do When GPU Temperature Is High
If your GPU is overheated, the temperature stays above your model’s safe range (around 85–90°C / 185–194°F on most NVIDIA cards) and the card starts to throttle, stutter or crash. Work through these steps in order, from easiest to hardest:
- Confirm the reading. Use HWiNFO or GPU-Z to check core, memory and (where available) hotspot temperatures while gaming, and make sure it is not a one-second spike.
- Check what else is using the GPU. Background apps, browser tabs with hardware acceleration and mining software can keep it busy even at idle.
- Clean the card and case. Remove dust from fans, heatsink fins and case filters.
- Improve airflow. Make sure there is clear front intake and rear/top exhaust, and give the card some space.
- Raise the fan curve. Make the GPU fans speed up earlier, for example from around 70°C (158°F).
- Lower the heat output. Undervolt, reduce the power limit or cap your FPS.
- Check thermal paste and pads. If the card is several years old, or the core-to-hotspot gap is above 30°C (54°F), repasting may help. Opening the cooler may void your warranty.
Stop using the card and get it checked if you see black screens, shutdowns, visual artifacts or fans that do not spin.
Best GPU Cooling Solutions
For most people, good case airflow and a clean air-cooled GPU are enough. Liquid cooling runs cooler and quieter on high-end builds, and immersion cooling is mostly for data centers.
Air Cooling (Fans)
Air cooling is standard on almost every card. Use front intake and rear/top exhaust fans, clean dust every few months and set a custom fan curve. Air-cooled cards typically sit around 70–80°C (158–176°F) under load.
Liquid Cooling
A radiator and pump move heat away faster. Liquid-cooled cards often run 55–65°C (131–149°F) under load, and one hybrid-kit test showed temperatures about 17°C (31°F) lower than stock air cooling. Cost and installation space are the downsides.
Immersion Cooling
The whole system sits in a non-conductive liquid. It removes heat extremely well but is used mainly in AI servers and data centers, not home PCs.
CPU vs GPU Temperature: PC Temperature Range
Your graphics card is not the only part that heats up. A safe CPU temperature is similar to a GPU’s under load, but CPUs and GPUs have different limits, so check each one separately. Here is the overall PC temperature range for a desktop:
| Component | Idle | Gaming / Heavy Load | Too Hot |
|---|---|---|---|
| CPU (desktop) | 30–50°C (86–122°F) | 60–85°C (140–185°F), under 80°C (176°F) is ideal | Sustained 85–90°C+ (185–194°F+) |
| GPU (desktop) | 30–50°C (86–122°F) | 65–85°C (149–185°F) | 85–90°C+ (185–194°F+), NVIDIA |
| Laptop CPU / GPU | 35–55°C (95–131°F) | 60–85°C (140–185°F), CPU can peak higher | Sustained 90°C+ (194°F+) |
A regular CPU temperature at idle is 30–50°C (86–122°F), and the average CPU temperature under gaming load sits between about 60°C and 80°C (140–176°F). The exact maximum differs by processor. In HWiNFO or your CPU’s spec page, look for Tj Max, the highest temperature the chip is designed to run at. Sustained readings close to it mean your cooling needs attention.
Two simple habits keep the whole PC healthy: keep both CPU and GPU under about 80°C (176°F) while gaming, and monitor them together in HWiNFO or MSI Afterburner.
Conclusion
A healthy GPU idles at 30–50°C (86–122°F), and gaming temperatures between 65°C and 85°C (149–185°F) are normal. Check your exact model’s limit, because it ranges from 85°C to 90°C on RTX 50 cards and up to 110°C (230°F) hotspot on Radeon. Monitor with HWiNFO, watch memory junction (and hotspot where your card reports it) as well as core temperature, and fix a high GPU temperature with cleaning, airflow, fan curves and undervolting. For more PC guides and tools, visit the Bottleneck Calculator blog.
FAQs
What is a good GPU temperature while gaming?
65–85°C (149–185°F) is normal. Below 80°C (176°F) is ideal.
What is a normal GPU temperature in Fahrenheit?
About 86–122°F at idle and 149–185°F while gaming. Above 185–194°F, most NVIDIA cards start to throttle.
Is 80°C too hot for a GPU?
No. 80°C (176°F) is normal under heavy load for most modern cards.
Is 90°C bad for a GPU?
90°C (194°F) is at or near the limit for NVIDIA cards. Short spikes are tolerated, but staying there causes throttling.
Why is my GPU overheating?
Common causes are dust, poor case airflow, a hot room, heavy settings, an overclock, or old thermal paste and pads. Check your temperatures in HWiNFO, clean the card and improve airflow first.
What is a good idle GPU temperature?
30–50°C (86–122°F). If it stays above 60°C (140°F), check background GPU usage, dust and airflow.
What is a normal hotspot temperature?
Usually 10–20°C (18–36°F) above the core on NVIDIA cards that report it. Some Radeon cards run 25–30°C (45–54°F) higher and still work normally. Hotspot up to 110°C (230°F) is within spec for Radeon. NVIDIA RTX 50 cards do not show hotspot in most standard tools.
Is 100°C memory junction temperature dangerous?
On GDDR6X cards, 100–110°C (212–230°F) is within design, but staying below about 95°C (203°F) is better. Sustained readings near the throttle point need better airflow or pads.
What is a normal laptop temperature while gaming?
Most gaming laptops run around 70–85°C (158–185°F) on the CPU and 60–85°C (140–185°F) on the GPU. Brief peaks above that can happen on thin models, but sustained 90°C+ (194°F+) means the cooling needs help.
What is a safe CPU temperature?
Most CPUs are comfortable under about 80°C (176°F) while gaming and 30–50°C (86–122°F) at idle. Check the Tj Max of your exact processor.
How do I lower GPU temperature?
Clean dust, improve case airflow, raise the fan curve, undervolt, cap FPS or upgrade the cooler.
Should I replace GPU thermal paste?
Only if the card is several years old, runs hot despite good airflow, or the core-to-hotspot gap is above 30°C (54°F). Opening the cooler may void your warranty.
Can I check my CPU temperature with the same tools?
Yes. HWiNFO shows CPU package and per-core temperatures right next to the GPU sensors, and MSI Afterburner can show CPU temperature in its in-game overlay.
