Heres' the reference card (Load = 66C / Load OC'd - 69C)
https://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_750_Ti/29.html
Heres' the Asus card (Load = 62C / Load OC'd - 64C)
https://www.techpowerup.com/reviews/ASUS/GTX_750_Ti_OC/29.html
So your are running way way high ... I'd proceed checking 1 at a time.
I don't recommend Furmark for stability testing but for testing cooling capability, its remains one of the more valuable tools a the load is constant. Asus, MSI and Gigglebyte all have recommended have each recommended we use it while diagnosing cooling issues as does Cleco for their laptops... Clevo recommends Furmark over OCCT . I'll run it on build day for 12 minutes or so to set a baseline... and then maybe every 3, 6 or 12 months depending on what I notice, then compare against baseline... if number is significant, I'll drain loop, clean blocks, flush sytem and reapply TIM. But unless its like 3C or more, won't bother... on build day it was 39C at 1250, 44c at current pump and fan speeds... if it hits even 50...55 no reason to really care, other than will have unrestful nerd sleep.
As long as nVidia is stating no worries till ya hit 95C, and card vendor TS is saying no problems, I'm gonan continue to think of the dire warnings as FUD. You need a utility that's is going to present a constant load, otherwise your initial baseline is meaningless. In any case, throttling should keep you below 85 even. That's not saying that just start the utility and then go make dinner. It's only 12 minutes with large liquid mass to heat up to a stabilized temp ... air cooled should be under 5. You should be able to sit there for 5 minutes and make sure that no scary temps are reached... just stop the test if ya see anything ya dont like. Again, If you break around 82C, the card should throttle to protect itself.
https://www.geforce.com/hardware/desktop-gpus/geforce-gtx-750-ti/specifications
Testing is usually done at 23C . Your in Saudi Arabia ...
- At 92F (33C) air temperature, that means +10 on the test results means you should see Load = 72C / Load OC'd - 74C
- At 110F (43C) air temperature, that means +10 on the test results means you should see Load = 82C / Load OC'd - 84C
So if that's the indoor air temps you are dealing with, your temps are perfectly normal.
Step 1 - Check Indoor Air Temp... take the test results from above and add (your temp in C - 23C) and add it to Load = 62C / Load OC'd - 64C. If you are 40C, then (40 - 23) + 62 or 79C should be your normal number. No sense in going any further if the case is ambient temps.
Step 2 - What is case cooling situation ? How many fans of which size, rpm, make and model ? The fact that you see temp changes with case open, tells me that case cooling is inadequate. If you have empty fan mounts, add fans... intake fans should = exhaust fans x 1.5 ... 2 out and 3 in works well and wont suck hot exhaust ai back inside case.
Step 3 - is air flow around case restricted in any way as in anything blocking air circulation around PC ? If so remove.
Step 4 ... I generally only need to do this in SLI builds as w/o good circulation around / between the cards, top card will be as much as 10C higher on top card. "Dead air" down around the card location can result in elevated temps... Many cases have screw holes allowing you to mount a fan on back of HD cage. These can work woinders in such situations... You can also use something like the Antec "SpotCool" to address such conditions.
Step 5 - You can try changing the paste ... but let's consider the fact that this is a 60 watt card. It's a good idea, but I think only after exhausting the above alternatives.