Just Google for liquid cooler benchmarks and look at the charts. Pick the best one you can afford and is available in your market.
Don't trust word of mouth opinions. Just look at actual tested numbers.

 

I've done that but the thing is that I'm seeing very mixed results. For example, Gamers Nexus's review shows that the LFIII maintains 80C avg with a 12900KS at 250W load, while Tom's Hardware's (where a 13700K is tested) says it throttles at 234W, and actually compares it directly with my current AK620, saying the following:

"If you look at their performance by total watts cooled, as shown in the chart below, you can see that the thermal performance of the two larger Liquid Freezers’ is similar to DeepCool’s AK620 air cooler."

My 13700K with AK620 starts hitting TJMax at around 210, and at 220W+ it's constantly throttling. I can maintain around 225-228W if I allow constant throttling, so that's almost in-line with TH's review (where it sustains 233W). GN's and TH's results are incredibly inconsistent, unless there are massive differences in the way the LFIII cools a 12900KS vs a 13700K, for some reason. This is why I though it'd be good to check with 13700K owners.

 

There isn't some magic that happens under the hood between the 13700 and 13900, or any chip really. Coolers don't do anything besides pull heat away from an IHS.
There's always going to be differences in testing environments, hence the discrepancies between different review sites.

If you're concerned, it's better to err on the side of caution, and expect worst-case scenario.

At the end of the day though, due to the nature of how air and water coolers work, the latter will always be a bit more efficient with cooling due to the water serving as a buffer.
Water doesn't instantly get hot. Hence, unless you're doing sustained loads to saturate all of the water, it will cool off quicker than the fins of an air cooler.

 

I mean, there are probably differences in cooling different CPUs due to the different heat density and heat spots.
That said, I was just reading this review:

ID-Cooling FX360 PRO Review - Shots Fired @ Arctic

ID-Cooling aims to make a statement with the FX360 PRO. The sleek and understated looks make for a great option if you prefer to keep things simple. This all-in-one liquid cooler delivers outstanding performance at a budget-friendly price of just $60, potentially earning it the "bang for the...

www.techpowerup.com

Here is says my AK620 can sustain 295W on a 13700K before hitting thermal throttle. And the LFIII 240mm (no 360 in the charts) can sustain 292W. Those are huuge differences across reviews, I have no idea what I can trust.

 

You can't compare cooling performance results from an AMD chip to an Intel one; not how it works. Your CPU is already tuned/optimized, so you can't compare those results to essentially any professional review since they test at stock settings only; they may also be in an open test bench which runs cooler than an enclosed case.

LF from Arctic is a great performer and cheap.

If you want something that is serviceable/upgradeable in the long run, then I'd suggest grabbing an Eisbaer Solo from Aquatuning; there's a few different models. This is the CPU block with pump/reservoir built-in allowing you grab a radiator(s) and fans of your choice.

The best benefit to this, outside of performance, is hoses that are exactly the right length and will look neat/tidy.

Spend a little extra up front for options later on. If at some point you want to watercool a GPU, you just expand your loop.

If you don't want to build it and fill the loop, they have pre-assembled and pre-filled kits that use G1/4 fittings you can alter if you so desire at some point.

 

Reviews are usually of sustained loads, not instantaneous ones.
So the fact still stands that water coolers are more efficient at cooling, as they have water as a buffer. Tower coolers do not.
Whether or not you deal in sustained loads depends on your own PC use-case, though.

Also, you can easily improve the performance of a water cooler by doing a push-pull configuration with two sets of fans instead of just one.
Tower coolers do not have that benefit.

Having personally used a Noctua NH-D15 (which ranks close to your DeepCool AK620) on several 12900K's back then, the experience was rather miserable.
It was very much a temporary cooler at best, to be inevitably changed after my overclocking was finished.

Go with a 360mm or 420mm AIO in push-pull. You won't go wrong. Resell that AK620.

 

Arctic dropped the ball completely on this cooler, by integrating its own anti-bend frame into it, requiring removal of the intel ILM. This means existing frames are useless, and that Arrow Lake compatibility is impossible (unless someone ends up testing that the exact IHS and socket dimensions are identical--not just the bolt spacings).
Arctic could have simply included their own add-on accessory contact frame, as well as a stock mechanism that would work on the original ILM and any aftermarket frames, but no--some people just don't like making lots of money.

 

Artic replaced a tedious mount for a janky one imo. Asus lists 1851 compatibility with most of it's higher end AIOs in specs of website cooler descriptions. Same for Artic. I hope the definition of compatibility isn't being reimagined.

Future-Proof Compatibility
The Liquid Freezer III is multi-compatible with Intel and AMD sockets. Intel will release the new Arrow Lake processors on the LGA1851 socket in the future. We guarantee full compatibility. All Liquid Freezer III can be used with the new LGA1851 socket from Intel without limitation.

 

That TomsHardware review is kind of an interesting can of worms. It got enough backlash to where the reviewer redesigned their methodology and retested the LF3. They got better results with the new methodology, but one thing you should be aware of is that he tests all coolers with a contact frame, which negates one of the key benefits for "average users" with the LF3. The reviews are useful for enthusiasts who would like to know how well the LF3 compares to other coolers when also using a contact frame, but it's not a good practice to only give those results because the overwhelming majority of users won't be using one. That is very relevant to your concerns @vg_vassilev - if you were or are planning to use a contact frame with a different cooler, then the TH review(s) may be informative for you, but if you are not, then they may be misleading.

Ultimately, I don't think the LF3 is the best cooler for Intel, simply because I don't trust the plastic contact frame and two-point mounting system. I would rather see a metal contact frame and 4-point mounting system. Their offset solution does make it one of the better ones for AMD, but that doesn't help you. It also has under-powered fans in my view, as I've seen tests with fan replacement where it does significantly better, even with just Arctic's own P12 Max fans, which I don't think would have raised the price very much.

 

You can still probably find LFIIs. I am running a 12700k with a LFII 420 version in push/pull. Not as effective as my other custom cooled rig, but pretty damned good for the money. I have built rigs for clients using the new version and I thought it was kinda nice, not to have to buy an extra brace for those builds and the cooling was good in all of my testing before they shipped out and have had O issues with any of them (13 and 14th gen i7s K series chips in all of em). I do prefer the LFII for myself. Cooling tests are tough to judge as everyone has different cases, fan setups and components. What works best on some testers bench (or test case) might not work best in your situation. Then again, I always use other fans with all my AiOs, and have custom fan curves. But Im not an average user either. I will spend the time running tests until its as fast as it will go, with as little noise possible.

 

I suggest EK-Nucleus 360

 

Hi all, sharing an update on my situation. I liked the EK-Nucleus 360 and wanted to get it, but I couldn't find it anywhere in stock. I got the Deepcool LT720 as I read its performance is very good, so I decided to give it a shot as I could always return it (spoiler, that's what happened). In R23 my CPU pulled 235W and in a 10-minute test the max temp with the LT720 reached around 97C, while I expected it to not exceed 90C. At 253W (OCCT CPU stress test) the package temp immediately spiked to 90C and after around 4 minutes it hits 100C and started throttling and bouncing off TJMax. I was quite disappointed with this, but the cherry on top was an annoying high-pitched sound the pump made, which I initially couldn't hear during the day (as I live next to a busy street), but once night came it was like I was standing right next to a refrigerator. And at full fan speed it's very loud overall, especially compared to my AK620. Interestingly, even at around 40% fan speed it was still noticeably louder than my AK620 at full blast. At 125W my AK620 maintained around 68-70C, while the LT720 - around 60C, but the AK620 is significantly quieter.

Today I returned the LT720, but it not being able to keep my CPU below TJMax at 253W seemed suspicious, so I concluded it is most likely due to poor contact with the IHS. So, upon returning the LT720, I bought a thermal grizzly contact frame. I just installed it two hours ago, and what can I say, it really makes a difference. Before switching the AK620 for the LT720 yesterday, I ran R23 at 200W PL2 with the AK620, and in just 2 minutes one P-core managed to thermal throttle at 95C (I typically have my TJMax set to 95C). Now, with the contact frame, the max temp at 200W is 89C, and this is at a 2-3C higher ambient temp (nice summer day today). The deltas in core temps across the cores are significantly improved, but also the overall max core temp.

Edit - an hour since I posted this comment, the ambient temp has now dropped to 22C and at this, at 200W the max temp in a 200W OCCT CPU stress test is 85C. This is a massive improvement compared to without the contact frame, somewhere around 10C.

With the contact frame I probably would've gotten significantly better results with the LT720, but even so, that pump noise (and loud fans) would've driven me crazy.
Nevertheless, now I know how annoying pump noise can be for me, so a quiet pump is my top priority, and if possible, quiet fans, but at least I could always replace those. As far as I've read the EK Nucleus CR360 should have a quiet pump, does anybody have one to share personal experience?

Another option I'm considering is getting an NH-D15 G2, as it would surely be some improvement over the AK620, there is no pump to worry about and it should be very quiet.

@Roboyto The more reviews I read the more I started to see how different AMD and Intel CPUs are to cool with the same cooler, indeed. Noted about the open test benches, unfortunately this is something that is not always specified in reviews. I decided not to go with the Arctic because I didn't want to bother with their custom contact frame, my thinking was that if I'm going to deal with a contact frame, it should allow me to use different coolers. Regarding this Eisbaer Solo, I haven't heard of that but I'll definitely check it out, thank you!

Reviews are usually of sustained loads, not instantaneous ones.

Yes, and this makes me even more confused because some of the results in some reviews just don't seem realistic. For example, TechPowerUp say the AK620 can sustain 290W before thermal throttling, which sounds crazy to me. Tom Hardware's review determines this threshold to be reached at 233W.

That TomsHardware review is kind of an interesting can of worms. It got enough backlash to where the reviewer redesigned their methodology and retested the LF3.

I saw that revised review too and from my POV, TH's methodology is the most "true to life". Their results in the AK620 review look pretty close to what I'm seeing. Looking at the LF3 revised results, they say it can only sustain 257W before the 14700K thermal throttles, while GN say the LF3 keeps their 12900KS at 80C at 250W (at a 2C lower ambient temp, but still). Those are incredibly big differences.

 

I saw that revised review too and from my POV, TH's methodology is the most "true to life". Their results in the AK620 review look pretty close to what I'm seeing. Looking at the LF3 revised results, they say it can only sustain 257W before the 14700K thermal throttles, while GN say the LF3 keeps their 12900KS at 80C at 250W (at a 2C lower ambient temp, but still). Those are incredibly big differences.

The "true to life" style results can provide an interesting data point, just like people's self-reported results on forums, but for me, the point of professional reviews is to show performance with as many variables removed as possible to give a "clean" perspective on relative performance between different coolers in like-for-like scenarios. The more you diverge from out of the box, open air performance, the more likely you are to represent an outlier than the norm, that's why all of the most trusted reviewers test on open air test benches with no special hardware (like a contact frame).