Why you can trust Tom's Hardware Our expert reviewers spend hours testing and comparing products and services so you can choose the best for you. Find out more about how we test.
Intel Core Ultra 9 285K Gaming Benchmarks — The TLDR
The particulars of our test setup are on page four. We used lifted power limits, which result in higher power consumption and heat, but you get faster performance in exchange. We revamped our Windows 11 test image to version 24H2 for this testing and updated all drivers and applications to the latest versions. We also added numerous new games to our test suite, using a mix of custom game scenes and built-in benchmarks.
Note that these game benchmarks aren't directly comparable with the results in our CPU benchmark hierarchy (we'll update as soon as we have enough chips to swap in the new test suite). We tested with Intel's Application Performance Optimization (APO) software active. This software will be installed by default via Windows Update, so it now represents default performance.
Here's the high-level view of gaming performance, using the geometric mean of our 14 gaming tests at 1080p. We're testing with an Nvidia GeForce RTX 4090 to reduce GPU-imposed bottlenecks as much as possible, and differences between test subjects will shrink with lesser cards or higher resolutions and fidelity. We have the game-by-game breakdowns further below.
We tested the Core Ultra 9 285K and Core Ultra 5 245K with a standard DDR5-7200 kit and a CUDIMM DDR5-8200 kit (marked as CU-8200 in the charts). Intel has made ambitious claims about power consumption during gaming, so we logged CPU power consumption and calculated the efficiency (FPS per watt) during each benchmark run.
The DDR5-8200 CUDIMMs only yielded an extra 2% of gaming performance over DDR5-7200, but we might see larger improvements as the platform matures. Some titles leverage the increased bandwidth better than others, as seen in the individual benchmarks below.
Even with CUDIMMs, the previous-gen Core i9-14900K is still 4% faster than the Core Ultra 9 285K, which falls short of Intel's claims of performance parity — but not by much. With both chips armed with DDR5-7200, the 14900K is 6% faster than the 285K.
Notably, the 285K ties the $599 Ryzen 9 9950X, which aligns with Intel's performance claims. Regardless, we don't like to see regressions in gen-on-gen performance, or even a lateral. Arrow Lake doesn't make forward progress in gaming, thus ceding the leadership position entirely to AMD's gaming-specialized Ryzen 7000X3D processors.
The $476 Ryzen 7 7800X3D is 16% faster than the 285K CUDIMM config and 19% faster than the 285K with DDR5-7200, cementing its spot as the fastest gaming CPU on the market. AMD also has its next-gen Ryzen 9000X3D chips coming to market, with the first slated to arrive early next month. As usual, you should take into account that X3D chips don't accelerate all games equally, and they deliver lower performance than standard models in productivity workloads.
The $309 Core Ultra 5 245K essentially offers the same overall gaming performance either with CUDIMMs or standard DDR5-7200 memory, with most of the individual benchmarks falling within the expected run-to-run variance or very close to it. At first blush, it appears the extra bandwidth from CUDIMMs isn't a huge benefit for this lower class of processor. However, we have much to learn about the new CUDIMMs moving forward, and we expect the platforms to become more refined over time, so this could change.
Regardless, the $256 Core i5-14600K is 5% faster in 1080p gaming than the $309 245K, which isn't great given the respective pricing. The plucky $209 Ryzen 7 5700X3D also ties the 245K in gaming, putting the 245K at even more of a disadvantage. The 5700X3D does come with the older Zen 3 architecture and slower clock speeds due to its stacked L3 cache, so it suffers tremendously in our productivity benchmarks. However, it snaps into the inexpensive and plentiful AM4 motherboard ecosystem, scoring a knockout win for budget gaming systems.
The $249 Ryzen 5 9600X is 6% faster than the Core Ultra 5 245K in gaming for $60 less, making for a potent adversary on the more modern AM5 platform. And with both platforms now requiring DDR5 memory, there's no potential savings to be had with DDR4 memory (unlike with the 14600K).
The Arrow Lake chips certainly don't paint a convincing picture for gamers chasing every last frame of performance, but they do consume less power during gaming than their predecessors and AMD's Zen 5-powered Ryzen 9000-series.
The 285K consumed an average of ~77W across our test suite, or about 36% less power than the 14900K. AMD's Ryzen 7000X3D processors consumed even less power than the 285K, though. However, the 285K consumes less power on average than AMD's Zen 5 Ryzen 9000 series during gaming.
Naturally, factoring in performance to generate our fps-per-watt metric is the best way to quantify power efficiency. Here, the 285K proves to be more efficient than the 9000 series, too. However, the Ryzen 7000X3D chips are even more efficient, with the 7800X3D running away with the win. Intel has made significant strides, though — the Core 9 285K delivers 52% more fps-per-watt than the prior-gen 14900K. That will inevitably lead to less thermal output.
We conducted a few basic temperature measurements for a general comparison, but our measurements weren't entirely equal. We used the ROG Ryujin III 360 ARGB Extreme 360mm AIO for the Core Ultra chips to ensure proper mounting pressure (Intel provided this cooler for the review), but all other test systems used the 360mm Corsair iCue Link H150i RGB (both coolers had fans and pumps at 100%). We also kept the room around 72F ambient, but temperatures aren't tightly controlled, so we don't have an entirely scientific measure.
With those factors in mind, we recorded the Core 9 285K running at an average of 48C across the test suite, while the previous-gen 14900K logged 60C. That's a 12-degree delta, generally aligning with Intel's claim of up to 10C lower temps with Arrow Lake. The 245K was also plenty impressive at 44C.
Arrow Lake isn't a homerun on the gaming front, leaving the top of the market to AMD's X3D chips while also leaving room for either AMD's Ryzen 9 9950X or Intel's own less expensive prior-gen counterpart, the Core i9-14900K, to step in as better all-rounders. The Core Ultra 9 285K does pack substantial gains in productivity performance that might change the equation for some, particularly if pricing becomes more forgiving.
The competition between AMD and Intel chips can vary based on the title (particularly with X3D models) and the GPU you use. It's best to make an informed decision based on the types of titles you frequently play, so be sure to check out the individual game tests below.
A Plague Tale: Requiem Benchmarks — Intel Core Ultra 9 285K
Baldur's Gate 3 Benchmarks — Intel Core Ultra 9 285K
Borderlands 3 Benchmarks — Intel Core Ultra 9 285K
The Ryzen 7000X3D chips take impressive leads in this title, but it doesn't represent Ryzen 7000X3D's performance in most titles. It also illustrates how outliers can make the X3D seem more impressive in cumulative measurements. It's worth noting that Borderlands 3 was an AMD promotional title, and that might play a role in its performance.
Cyberpunk 2077 Benchmarks — Intel Core Ultra 9 285K
Our Cyberpunk 2077 results are suspiciously odd, especially because the 285K lags behind the 245K. We're working to diagnose this issue but have confirmed this result with multiple retests. Our GPU tester also noticed a lot of anomalies with the latest 2.13 patch that came out last month (and added FSR3.1 support), with inexplicable dips in performance.
F1 2024 Benchmarks — Intel Core Ultra 9 285K
Far Cry 6 Benchmarks — Intel Core Ultra 9 285K
Far Cry 6 is another AMD-promoted game, and the 7800X3D truly excels thanks to its single-CCD design paired with its extra slab of L3 cache.
Final Fantasy XIV Benchmarks — Intel Core Ultra 9 285K
Hitman 3 Benchmarks — Intel Core Ultra 9 285K
Hitman 3 leverages Intel's E-cores for certain game functions, but it also likes Ryzen X3D's voluminous L3 cache.
Hogwarts Legacy Benchmarks — Intel Core Ultra 9 285K
Microsoft Flight Simulator Benchmarks — Intel Core Ultra 9 285K
Microsoft Flight Simulator 2021 obviously benefits from L3 cache — the Ryzen 7000X3D chips are incredible in this title.
Minecraft Benchmarks — Intel Core Ultra 9 285K
Minecraft can be taxing on the CPU in some scenarios, and we've devised a CPU benchmark by leveraging the built-in test in the Portal Pioneers RTX ray-tracing benchmark. However, we disable ray-tracing to limit the GPU overhead during this test and dial up the render distance to the maximum 95 chunks. Then, we simply hop into the minecart and take a ride around the track, which gives us a nice repeatable scene for benchmarking, albeit one that is more taxing than more mundane settings.
Spider-Man: Remastered Benchmarks — Intel Core Ultra 9 285K
Spider-Man represents one of the few high points for Arrow Lake in gaming performance. It has curious rankings in general, however, with the 7950X3D taking the top spot, and even the 7900X3D matches the 7800X3D in performance. It seems that the game appreciates CPU throughput at least as much as having large amounts of cache. Also curious is that Raptor Lake didn't do so well, and the 14900K result may need further investigation as it came out below both the 14600K and the 12900K. But the 13900K also did poorly, so it seems that in general the game doesn't appreciate the specific E-core/P-core split on certain Raptor Lake models.
Starfield Benchmarks — Intel Core Ultra 9 285K
Here, finally, is the one game we've tested where Arrow Lake gets a win. And it's a fairly sizeable win considering it has trailed in many other games by 5–10 percent, so coming out up to 7% ahead of the 7800X3D represents a big shift. Starfield seems to care less about memory latency and more about CPU throughput, but this tends to be the exception (for gaming) rather than the rule.
Watch Dogs Legion Benchmarks — Intel Core Ultra 9 285K
Wrapping up and summarizing the gaming results, Watch Dogs Legion has the 285K just ahead of the 14900K, with CUDIMMs providing a 4% improvement in performance on the 285K. With regular DDR5 memory, the old and new Intel chips are effectively tied. AMD's 7950X3D and 7800X3D still take the top two spots, while the 7900X3D falls well off the pace set by its siblings.
Overall, Arrow Lake fails to impress with its gaming performance. If you're not using a top-tier GPU, the differences are going to be a lot less noticeable, but any regression at all in performance can only be seen as a disappointment. If you want the fastest gaming CPU, that remains the Ryzen 7 7800X3D for now, though we anticipate at least a moderate bump in performance when the presumed Ryzen 7 9800X3D arrives on November 7.
