Monday, February 2nd 2026
Intel's Core Ultra 9 290K Plus Shows 10% Performance Boost Over Core Ultra 9 285K
Intel's next desktop "Arrow Lake Refresh" CPU upgrade is inching closer to reality. The Core Ultra 9 290K Plus has been spotted in a new Geekbench run, adding to the evidence that the "Arrow Lake Refresh" will indeed offer a meaningful performance improvements over its predecessors. The test system used an ASUS ROG Strix Z890-E Gaming Wi-Fi board with 64 GB of DDR5-6800 memory. The processor achieved scores of 3,535 in the single-core test and 25,106 in the multicore test. Compared to the Core Ultra 9 285K's typical scores of around 3,200 and 22,560, this represents improvements of approximately 10.5% and 11.3%, respectively. These results place the 290K Plus at the top of Intel's consumer CPU rankings in Geekbench's database. An earlier leak on different hardware showed slightly lower results, suggesting that this newer test run benefits from better optimization rather than just faster memory.
The 290K Plus SKU keeps the same 24-core layout as its predecessor with 8 P-Cores and 16 E-Cores, plus identical power limits of PL1 of 125 W and PL2 of 250 W. The gains come from higher clock speeds. According to rumors, the efficiency cores now boost to 4.8 GHz, up 200 MHz, while the performance cores get an extra 100 MHz on both turbo and thermal velocity boost. The benchmark registered the chip running at 5.7 GHz during testing. Intel has confirmed the ARLR is coming but has stayed quiet on specific models and dates. Leaks suggest a March or April release, and since these chips use the same LGA 1851 socket, they should work as drop-in upgrades for current Z890 motherboards. As with any pre-release numbers, may not reflect final CPU performance, so final performance data and gaming results that come from third-party reviews will show the real-world situation.
Sources:
Geekbench v6, via VideoCardz
The 290K Plus SKU keeps the same 24-core layout as its predecessor with 8 P-Cores and 16 E-Cores, plus identical power limits of PL1 of 125 W and PL2 of 250 W. The gains come from higher clock speeds. According to rumors, the efficiency cores now boost to 4.8 GHz, up 200 MHz, while the performance cores get an extra 100 MHz on both turbo and thermal velocity boost. The benchmark registered the chip running at 5.7 GHz during testing. Intel has confirmed the ARLR is coming but has stayed quiet on specific models and dates. Leaks suggest a March or April release, and since these chips use the same LGA 1851 socket, they should work as drop-in upgrades for current Z890 motherboards. As with any pre-release numbers, may not reflect final CPU performance, so final performance data and gaming results that come from third-party reviews will show the real-world situation.
32 Comments on Intel's Core Ultra 9 290K Plus Shows 10% Performance Boost Over Core Ultra 9 285K
Then there is still the problem with power draw... These 200 Watts + chips needs monster cooling to keep it somewhat acceptable temperature.
if you are a gamer and don't play at 4K get an x3d chip
If you use software that benefits from the large victim cache get the x3d chip
for all the rest it's stupid close and if the 1% difference is that significant to you getting your job done faster it's time to look in to getting a threadripper/xeon system
I would buy your boost frequency theory for multi-core via binning / process node improvements and so on, but single core +10.5% from a +2% clock speed boost implies there is more to it. Not sure I buy the tightened timings either, IIRC MT Geekbench was very latency sensitive but ST was not.
But if perchance it is tightened up timings, well that was supposedly what hurt gaming performance.
Regardless of the reason, this is clearly not just a clock bump.
I have a suspicion the reviews will be very interesting.
MT:
I did find a heavily OC'd 285K on a Taichi that could handily beat the 290K baseline in MT, but interestingly it was still ~1% behind in ST.
This is basically like having a stable 6.1Ghz 285K
I already maxed out mine with, 5.7P 5.1E ,42 Ring, 40 D2D, and 35 NGU with 9000MT's cl38 RAM, but I will gamble my luck on these refresh chips somehow.
Seems like they are comparing a low 285K score to a high 290K score.
Keep in mind these are not OC results, for either system.
Simple arithmetic indicates 290K is not 10% faster.
Notice how 285K is only 1.5% behind 9950X, it's regular non-X3D competitor...
But if you're talking about everything else that is not 720p, including productivity, uh no. Arrow Lake is competitive, and that's before any real tuning.
While being harder to cool because of their CCD architecture, and bad IHS. The only "monster cooling" you really need is a delid, because the real bottleneck is getting the heat out of the silicon fast, not how much rads you have.
Modern chips are very efficient, but if you need even more and don't like seeing 200W of power usage, just power limit it.
If the normal Arrow Lake chips get more price cuts after this boring refresh, that's good.
Oh boy, I smell doubt on your post..
This same kit runs 8400 cl36 as well on my Z790 Apex Encore and a binned 13900KS, kit gets maxed out, same chip does 8600MT's cl38 on a pair of Patriot Xtremes
If the x86 cores and the memory controller were on the same die, made with the most advanced lithography available, the memory controller would heat up less and the CPU would likely support higher clock speed RAM more easily.
What should be placed on the same die as the x86 cores:
- PCIe controller
- Memory controller
- iGPU (RDNA5!)
All other components could be placed on a separate die, made with cheaper lithography.
Ignoring what consumers want is a sure way to lose sales and market share. Intel did this for many years and is now reaping what it sowed.