Apparently it is running more efficiently than the GMKtec Evo-X2. This might explain why it was getting better results in several tests. Though I suspect some tests were done on a balanced power profile which will only limit it to 100w.
“the Framework system draws an average of ~144 W when running Prime95 compared to ~186 W on the GMK”
Hmmm, but it seems they did put the fan the wrong way, wouldn’t that affect the cooling performances? 
EDIT: it seems they tested both directions:
(September 18, 2025 update: CPU fan results in both orientations will be added soon!)
In the prime95 stress test they got a average of 64c so I’m not sure if they had the fan on the correct way for that test but its seems a single fan is able to keep it cool but it depends on your ambient temperature. I would be interested in a mini-itx case temperature test with like 2 or 3 case fans.
No matter what they say they used, the power draw they measured suggests they used “balanced” power settings, at least that is the power draw I see on Linux with those settings. I don’t know if Windows is maybe handling those power modes differently in some way.
The Framework can use more power than that. If you set it on Linux to “Performance” sustained power draw is 20W higher and you may get a few % points more in performance.
