Arctic MX-6 vs MX-4

A year and a half ago I compared Arctic’s MX-5 thermal paste to their extremely popular MX-4 paste. MX-4 remained my thermal paste of choice despite a half-degree temperature drop with the use of MX-5, because it is very easy to apply due to low viscosity. MX-5 certainly performed well, but it was very viscous and sticky in comparison to MX-4. Arctic has now discontinued MX-5, apparently due to issues with consistency, and they’ve recently launched a new thermal paste, with MX-6. Just like with MX-5, I decided to give MX-6 a try and compare it to MX-4 side-by-side to see if it offers any improvement. Based on the technical specifications, MX-6 has a higher viscosity and higher density than MX-4. They both appear very similar, with a dull grey color. I prefer to manually spread thermal paste across a CPU’s surface to make sure that it coats the entire heat spreader. Be careful with Intel’s 12th and 13th generation CPUs however. They can be bent due to the force applied by the stock locking mechanism, leading to less-than-ideal contact with the cooler in the middle of the CPU. I use bending correction frames like this one from Thermalright to prevent the CPU from bending. It is clear that the MX-6 paste is more viscous, and it does not spread as easily as MX-4, but it is better than MX-5 in this regard. For my test setup, I am using an open-air PC frame, called the Monolith from Cooltech. Cooltech sent this PC frame to me for review, and I’ll do a separate video the Monolith later. This PC included the Intel Core i9-12900K CPU, Asus ROG Strix B660-I Gaming WiFi motherboard, Noctua NH-P1 CPU cooler, and an Intel Arc A380 graphics card with a fanless Arctic Accelero S1 GPU cooler. I used half-hour Prime95 Torture Tests to stress the system and HWInfo to collect thermal data. The CPU was tested at 15-watt power limit intervals, between 35 watts and 125 watts. I performed the exact same tests with MX-4, MX-6, and MX-5 for comparison. The data I am presenting here are the maximum 1-minute average core temperatures at the end of each half-hour stress test. The ambient room temperature was maintained at between 17.5 and 18 degrees Celsius. The room temperature was tracked and subtracted from the CPU temperatures to equalize results. For the Intel i9 CPU, temperatures were slightly lower overall with the MX-5 and MX-6 pastes. The average temperature drop from MX-4 was 0.3 degrees for MX-5 and 0.2 degrees for MX-6, but there was something interesting here. MX-4 seemed to perform better at the lower-power end of the data, and MX-5 and MX-6 seemed to perform better at the higher end, where it really matters. So, despite just a 0.2- or overall temperature drop, at above 100 watts, the average temperature drop from MX-4 was 0.7 degrees for MX-5 and 1.2 degrees for MX-6. If I look at this data a different way, and use trendlines to try to estimate the power limit at which I would expect to see thermal throttling at 100 degrees, in a 25-degree room, I get 132 watts for MX-4, 133 watts for MX-5, and 134 watts for MX-6. It’s certainly not a huge difference, but Arctic’s MX-6 does look like it offers a benefit when compared to their MX-4 paste. The benefit is potentially a little more than one-degree lower temperatures. Like MX-5, MX-6 is not as fluid as MX-4, but it doesn’t have the stickiness of MX-5. The higher viscosity may help in some situations, such as with direct-die cooling. On Amazon in the US, at the time of this video, 2 grams of MX-4 costs 4 dollars 79 cents, and 2 grams of MX-6 costs 6 dollars 79 cents. 4 grams of MX-4 costs 5 dollars 30 cents, and 4 grams of MX-6 costs 8 dollars 49 cents. 8 grams of MX-4 costs 6 dollars 65 cents, and 8 grams of MX-6 costs 11 dollars 99 cents. Personally, wouldn’t consider anything less than one full degree lower to be significant, and while MX-5 didn’t quite reach that, MX-6 did manage to end up at more than one degree lower than MX-4 at the high end. The fluidness of MX-4 still draws me to MX-4, but I think I’ll start keeping some MX-6 in my workshop as well. To get photos and test results of every one of my PC builds, check out my Patreon page. Like the video and subscribe for more fanless PC content, and visit in you are interested in purchasing your own custom-built fanless PC.