MSI Crosshair 18 HX AI review

Frame rates

1920×1080, High preset, upscaling off. Averages 141fps across the 5 core benchmark games.

Measured on this machine. Source: Notebookcheck, 2025-08-30.

In use

The short version

The MSI Crosshair 18 HX AI is a very good gaming laptop wrapped around a genuinely excellent screen, sold at a price that makes it hard to recommend to most people. Everything about how it performs is defensible. Everything about what it costs is not.

Start with what works. Notebookcheck's measurements at 1920x1080 on High presets with upscaling off put this machine at the upper end of what an RTX 5070 laptop delivers. The five games in our scoring suite average 141fps, and the spread from 280fps in Strange Brigade down to 62fps in Black Myth: Wukong is the normal shape for this class of hardware rather than evidence of anything going wrong. Against the four other RTX 5070 machines in our database that share test titles with it, this one wins three of the four comparisons. The 18in 2560x1600 240Hz IPS panel is the best component in the machine by some distance — the resolution is right for the size, the refresh rate is high enough to matter in the games that can reach it, and 18in of 1600p is a genuinely different working surface from the 16in panels most of its rivals use.

Now the problem. It costs £2,095, and what you get for that is an RTX 5070 with 8GB of video memory. That is the specific combination that makes this segment of the market so unrewarding right now: enough GPU to push high frame rates today, not enough memory to be confident about texture settings in two years, and a price that sits close enough to genuinely faster hardware to invite awkward comparisons. Nothing in the benchmark data is disappointing. The disappointment is entirely in the relationship between those numbers and the number on the price tag.

It also weighs 3.116kg before you add a charger, and it is an 18in machine. This is not a laptop in any meaningful behavioural sense. It is a small, awkward desktop that folds shut. If your plan involves carrying it anywhere on a regular basis, stop reading and look at something two screen sizes down.

Buy it if you want the biggest, best-specified panel you can get with a mid-tier GPU behind it, you have a desk to put it on, and you have made peace with paying a premium for MSI's 18in chassis. Do not buy it if you are shopping on frames per pound, if you want to play at the panel's native resolution with everything turned up in the most demanding new releases, or if the machine needs to leave the house more than occasionally.

Performance

What the measured figures actually cover

Every frame rate in this review comes from Notebookcheck's measurements dated 30 August 2025, taken at 1920x1080 on High presets with upscaling off. That last detail matters more than people usually give it credit for. These are native-rendered numbers with no DLSS, no frame generation, no upscaling of any kind. They represent the raw output of the GPU rather than what the machine will show you once you enable the features NVIDIA expects you to enable. In practice, most owners will turn DLSS on in the heavier titles and see higher numbers than the ones below. The value of measuring without it is that it gives you a clean, comparable picture of the silicon rather than a picture of how aggressively a particular game's upscaler has been tuned.

The other thing to be clear about: 1920x1080 is not this machine's native resolution. The panel is 2560x1600, which is roughly 2.1 times as many pixels. We do not have measurements at native resolution, and we are not going to guess at them, because the scaling from 1080p to 1600p is not a fixed ratio — it depends heavily on how GPU-limited a given title is, how much of the frame time is spent on resolution-independent work, and whether video memory becomes a constraint at the higher resolution. What we can say confidently is that the figures below are the ceiling, not the expected experience at the resolution you actually own. Every number here should be read as "this is what the GPU can do when it is not being asked to fill the whole panel".

That is not a criticism of the testing methodology. 1080p High is a sensible common denominator, and it is what makes cross-machine comparison possible in the first place. But it does mean the gap between these figures and native-resolution reality is larger on an 18in 1600p machine than it would be on a 1080p one, and you should hold that thought through the whole of this section.

The top end: Strange Brigade and F1 24

Strange Brigade at 280fps is the highest figure in the set and the least informative. Strange Brigade is a Vulkan title that scales beautifully, has modest geometric complexity by modern standards, and has been a benchmark staple precisely because it stresses raw throughput without much else getting in the way. A number this high tells you the GPU is not being throttled into the ground and the CPU is not creating a bottleneck at 1080p. It does not tell you much about what any recent game will do. Treat it as a diagnostic rather than a promise.

F1 24 at 223fps is more interesting. Racing games are one of the few genres where very high refresh rates translate into something you can genuinely feel — the input-to-photon latency reduction is real when you are making constant small steering corrections, and the visual coherence of a track flowing past you at 200+fps is noticeably cleaner than at 100. This figure sits just under the panel's 240Hz ceiling, which means F1 24 at 1080p High is effectively a refresh-rate-limited experience on this machine. That is a good place to be. At native 1600p it will not be, but F1 24 is also one of the titles where turning on DLSS Quality costs you almost nothing visually, so this is a realistic candidate for a high-refresh experience at the panel's real resolution.

These two results together establish the ceiling: when nothing is in the way, this GPU produces a lot of frames. That is worth knowing before we look at the games where something is in the way.

The middle band: Dota 2, GTA V, Final Fantasy XV, Baldur's Gate 3

Dota 2 at 182fps and GTA V at 177fps are the two oldest titles in the scoring suite and they behave accordingly. Dota 2 in particular is far more sensitive to CPU single-thread performance and driver overhead than it is to GPU horsepower, and 182fps is a healthy figure that says the Core Ultra 9 275HX is doing its job. GTA V is a decade-old game with a rendering pipeline designed for hardware that no longer exists; at 1080p High it is essentially a CPU and memory-bandwidth test. Both of these being comfortably above 150fps means competitive and older titles are a non-issue on this machine, which matters if a meaningful share of your playtime goes into things like CS2, Valorant, League, or older single-player games that you never got round to.

Final Fantasy XV at 145fps sits in the awkward middle. It is a heavier engine than GTA V, with more expensive post-processing and denser scenes, and the drop of over 30fps from GTA V reflects that. It is still well above the 120fps mark, which is where high-refresh gaming stops being a rounding error and starts being visibly smoother than 60.

Baldur's Gate 3 at 133fps is one of the more useful numbers in the whole set, because BG3 is a modern game that a very large number of people actually play, and because its performance profile is unusual — it is heavily CPU-dependent in busy areas like Act 3's city, and heavily GPU-dependent in the graphically dense outdoor environments. A 133fps average at 1080p High means the machine has enough headroom in both directions that neither becomes a problem. This is also the sort of game where the 18in panel earns its money: BG3's interface is dense with small text and small icons, and more physical screen area at a high resolution makes it substantially more comfortable to play than it is on a 15in laptop.

Cyberpunk 2077 and X-Plane 11: where the work gets real

Cyberpunk 2077 at 115fps is, for our purposes, the single most representative number here. Cyberpunk's High preset without ray tracing is a demanding but not punishing workload, and it has become the industry's default yardstick for a reason: it scales sensibly, it stresses the whole pipeline, and it looks like the kind of game people buy machines like this for. 115fps at 1080p High is a good result for an RTX 5070. It is comfortably above the 100fps threshold that lets a 240Hz panel do something useful, and it leaves genuine headroom for turning individual settings up rather than having to negotiate downwards.

The caveat is the one we flagged earlier. Cyberpunk at 2560x1600 native, no upscaling, High preset will not be at 115fps. It will be meaningfully lower, and we do not have the measurement to tell you exactly how much lower. What we can say is that Cyberpunk is one of the best-implemented DLSS titles on the market, and DLSS Quality at 1600p renders at a resolution close enough to 1080p that the measured figure becomes a reasonable proxy for the upscaled native-panel experience. If you are buying this machine to play Cyberpunk-class games on its own screen, you are buying an upscaling machine. That is not a scandal — almost everything at this tier is — but it should be a conscious decision rather than a surprise.

X-Plane 11 at 100fps is the outlier of the scoring suite and deserves an explanation, because on paper 100fps looks like the GPU struggling and it is nothing of the sort. X-Plane 11 is an old-generation OpenGL flight simulator with a rendering approach that leans extremely hard on single-threaded CPU work and draw-call submission. It is one of the very few titles in common benchmark suites where a faster GPU buys you almost nothing and a faster CPU buys you a lot. A figure of 100fps here says more about the Core Ultra 9 275HX and the platform's memory subsystem than it does about the RTX 5070. If you fly sims, the relevant conclusion is that this machine's CPU is strong — which it is, this is one of the fastest mobile chips available — and that the result would look similar with a considerably more expensive GPU fitted.

The presence of X-Plane in the scoring average also drags the mean down in a way that slightly undersells the GPU. The 141fps figure across the five scoring titles is honest, but two of those five (Dota 2 and X-Plane 11) are substantially CPU-bound, which means the average is a blend of graphics performance and platform performance rather than a pure GPU reading. That is deliberate — real gaming is a blend too — but it is worth understanding when you compare against machines with weaker CPUs.

Black Myth: Wukong at 62fps, and what the floor tells you

Black Myth: Wukong at 62fps is the lowest measured result and the most important one for anybody deciding whether this machine will still be adequate in three years. Wukong is an Unreal Engine 5 title with heavy use of Lumen global illumination and Nanite geometry, and it has become the de facto stress test for the current generation because it is genuinely, unapologetically expensive to render. 62fps at 1080p High with no upscaling is a respectable result for an RTX 5070 — it is above 60, it is playable, and it is not a disaster.

It is also the shape of things to come. UE5 with Lumen is where a large fraction of the next few years of AAA development is heading, and Wukong is a reasonable preview of what those games will demand. The gap between Wukong's 62fps and Cyberpunk's 115fps is the gap between the previous rendering generation and the current one, and it is enormous — Wukong runs at roughly 54% of Cyberpunk's frame rate on identical hardware at identical settings. Extrapolate that forward and the honest conclusion is that this machine will handle UE5 titles at 1080p-class internal resolutions with upscaling for the life of a typical ownership period, and will not handle them at 1600p native. Again: that is normal for this GPU tier. But at £2,095 it is worth stating plainly.

The 62fps figure is also the number that makes the case against the 240Hz panel being the deciding feature. In the games where high refresh rates matter most — competitive shooters, racers, older titles — the GPU reaches or approaches the panel's capability. In the games where the panel's resolution and size matter most — big, dense, beautiful single-player worlds — the GPU is producing frames at a rate a 120Hz panel would handle without complaint. The two halves of the machine are optimised for different things.

The GPU power limit

The fact sheet does not give us a confirmed TGP figure for the RTX 5070 in this chassis, and we are not going to state one we cannot source. What we can do is reason from the comparative data, which is unusually informative here.

The RTX 5070 laptop GPU is sold across a wide power band, and the difference between a low-power implementation and a high-power one is not marginal — it is routinely the difference between two nominally identical machines performing 15-20% apart. Our database contains a Razer Blade 14 running the same RTX 5070 at a confirmed 115W, which sits toward the lower end of the range because it is a 14in chassis with the thermal budget that implies. The Crosshair 18 HX AI averages +12fps against that machine across 8 shared games.

A 12fps average advantage across eight titles, on identical GPU silicon and with a strong CPU on both sides, is consistent with this machine running its GPU meaningfully above 115W. It is not proof — chassis cooling, memory clocks and driver-level differences all contribute — but combined with the absolute figures (280fps in Strange Brigade, 223fps in F1 24, 115fps in Cyberpunk) it points fairly clearly at an implementation near the top of the RTX 5070's range rather than the bottom.

Why this matters practically: it means you are getting the good version of this GPU. When you see an RTX 5070 laptop advertised, the badge tells you very little on its own, and a substantial fraction of the machines carrying it will be slower than this one by a margin you would notice. The 18in chassis exists partly to make that thermal headroom possible, and the benchmark data suggests it is being used. If you are going to pay a premium for a large machine, extracting full performance from the GPU is one of the better reasons to do it.

What we cannot tell you is what that costs in fan noise or surface temperature, because we do not have measured acoustic or thermal figures for this machine. Large chassis generally run quieter than small ones at equivalent power, and MSI's 18in machines have historically had substantial cooling, but that is a general expectation rather than a measurement and we will not present it as one.

Eight gigabytes of video memory

This is the specification that should give a prospective buyer the most pause, and it is worth being precise about why.

The RTX 5070 in this machine has 8GB of VRAM. At 1080p, in the games measured, that is sufficient — none of the results above look like they are being constrained by memory capacity, and a memory-starved GPU produces a characteristic pattern of catastrophic frame time spikes rather than a merely lower average, which is not what the data shows.

The concern is not today. It is the interaction between three things: the panel is 2560x1600, which increases framebuffer and render-target memory consumption; texture budgets in new releases have been trending upward for several years; and the ownership period for a £2,095 machine is realistically four or five years, not two. Video memory is the one component in a laptop that cannot be upgraded, worked around, or compensated for with settings adjustments that do not visibly degrade the image. When you run out of it, you turn textures down, and turning textures down is the single most visually obvious concession available in a modern graphics menu.

Put in blunt terms: the GPU in this machine will still be producing acceptable frame rates in demanding titles for years, and the memory attached to it may force you to make the specific compromise that most degrades how those titles look. That is a structurally awkward pairing. It is also, to be fair, not unique to this machine — the 8GB RTX 5070 is what the market is currently selling — but the more you pay, the less tolerable it becomes. At £1,400 you would shrug. At £2,095 it is a legitimate grievance.

The 32GB of system memory partially offsets this in the sense that texture streaming from system RAM is less painful than streaming from an SSD, and 32GB is the right amount for a machine at this price. The 1TB SSD is adequate rather than generous; three or four modern AAA installations will fill it, and you will want to plan for either external storage or a second drive if the chassis allows it.

The CPU

The Core Ultra 9 275HX is a genuinely fast processor and it is doing visible work in this benchmark set. The Dota 2, GTA V and X-Plane 11 results are all substantially CPU-determined, and all three are strong. This has two practical consequences.

First, if your gaming includes simulation, strategy, or heavily-modded titles — the categories where CPU performance dominates — this machine will handle them better than its GPU tier would suggest. Cities, Total War campaign turns, flight sims, large-scale strategy: these are workloads where the 275HX matters more than the difference between a 5070 and a 5080.

Second, if you do serious non-gaming work — compilation, video encoding, 3D scenes, large datasets — the CPU is a substantial part of what you are paying for and it is a good one. That reframes the value question somewhat. As a pure gaming purchase, £2,095 for these frame rates is a poor deal. As a workstation that also games, with a flagship-tier mobile CPU, 32GB of memory and an 18in high-resolution panel, the arithmetic is more defensible. We score machines on gaming, because that is what this site is for, but if half your use is professional the calculation genuinely changes.

Reading the spread as a whole

The distance from 280fps to 62fps is a 4.5x range across nine titles, and that spread is the most honest summary of the machine's character. It is not a machine with a single performance level; it is a machine whose behaviour is entirely determined by what you run on it.

Broadly, three tiers emerge. Competitive and older titles (Strange Brigade, F1 24, Dota 2, GTA V) run at 175fps and above, comfortably into high-refresh territory, and would still do so with the resolution increased. Modern but conventionally-rendered games (Final Fantasy XV, Baldur's Gate 3, Cyberpunk 2077) land between 115 and 145fps at 1080p, which becomes a solid 60-100fps-class experience at native resolution with upscaling engaged. Current-generation UE5 titles (Black Myth: Wukong) sit at 62fps at 1080p, which becomes a 60fps-target experience at native resolution with upscaling, and probably a 60fps-target experience with some settings reduced as engines get heavier.

None of those three tiers is bad. The first is excellent, the second is good, the third is adequate. The question the rest of this review has to answer is whether that stack of outcomes is worth £2,095, and the short answer is that it is worth it only if the panel and the chassis are specifically what you want.

The screen

The panel itself

An 18in 2560x1600 240Hz IPS panel is, on paper, close to the best configuration available in a gaming laptop that is not an OLED. Each of those four specifications is the right choice, and the combination is rarer than it should be.

The resolution is correct for the size. 2560x1600 across 18 inches gives a pixel density in the region where individual pixels stop being distracting at normal viewing distance without pushing scaling requirements to the point where Windows becomes awkward. It is also the resolution at which a mid-tier GPU can plausibly operate — 4K on a laptop with an RTX 5070 would be an exercise in permanent compromise, and 1080p across 18 inches would look coarse. 1600p at 18in is the sensible middle, and it is a substantially better working surface than the 1600p-at-16in panels that most of this machine's rivals use, simply because it is physically larger.

The 16:10 aspect ratio that 2560x1600 implies is worth a word. The extra vertical space is a real benefit outside games — more visible lines of code, more of a document, more timeline in a video editor — and inside games it is at worst neutral and often mildly advantageous. There is no downside here beyond letterboxing on 16:9 video, which is a triviality.

The 240Hz refresh rate is genuinely useful in a subset of what you will run. The measured figures show four of the nine tested titles above 175fps at 1080p, and two of those (Strange Brigade at 280fps, F1 24 at 223fps) are in the range where a 240Hz panel is being properly exercised. If you play competitive titles — and the CPU and GPU here are more than capable in that category — the refresh rate is not marketing, it is the reason to buy this panel over a 165Hz one.

IPS is the pragmatic choice. It gives you viewing angles that hold up when the machine is on a desk rather than directly in front of your face, colour behaviour that is predictable, and no risk of burn-in from static interface elements — which matters more than people admit on a machine that will spend hours a day showing the same taskbar and the same game HUD. What it does not give you is OLED's black level or per-pixel response. An OLED panel at this size and refresh rate would look better in a dark room playing a dark game, and would cost more. We do not have measured contrast, brightness or colour gamut figures for this specific panel, so we are not going to characterise its image quality beyond what the panel type implies.

Whether the GPU can drive it

This is where the machine's internal tension lives, and it is worth working through carefully because it determines what the panel is actually for.

The measurements are at 1080p. The panel is 1600p. Approximately 2.1 times the pixels. In the titles at the top of the range, that headroom exists — Strange Brigade at 280fps and F1 24 at 223fps have enough margin that even a substantial native-resolution penalty leaves them well into high-refresh territory. Dota 2 at 182fps and GTA V at 177fps are largely CPU-limited, which means the resolution increase costs them proportionally less than it would a GPU-limited title. For this whole category, the panel is a straightforward asset: high refresh, high resolution, and enough GPU to feed it.

In the middle band it gets more interesting. Baldur's Gate 3 at 133fps and Cyberpunk 2077 at 115fps have real but finite headroom. Without measurements at native resolution we will not put a number on where they land, but the direction is unambiguous and the magnitude is substantial. In practice these are DLSS titles at 1600p on this machine — set the upscaler to Quality, render internally at something close to the resolution these figures were measured at, and you get a native-panel experience at a frame rate in the neighbourhood of the measured numbers. That is a perfectly good outcome. It is also not the same thing as "the GPU drives the panel natively", and anyone shopping on the basis that 2560x1600 native, no upscaling, High settings is the target should look at more expensive hardware.

At the bottom, Black Myth: Wukong's 62fps at 1080p makes the position clear. This is a title where the panel's resolution is a liability rather than an asset for an RTX 5070, and where the 240Hz refresh rate is entirely academic. You will run it upscaled, targeting 60fps, and the panel will be doing nothing that a cheaper 120Hz screen could not.

The honest summary: the panel is excellent, and the GPU behind it is one tier short of matching it in the most demanding games. That mismatch is not a flaw in either component — it is a consequence of MSI fitting a flagship-class display to a mid-tier graphics configuration, which is a defensible product decision but one that shifts the value proposition toward people who care about the screen for reasons beyond frame rate. If a large share of your time on this machine is spent on the desktop, in a browser, in a document, or in an application rather than a game, that 18in 1600p panel is worth a great deal. If every hour is spent in Unreal Engine 5 titles, it is worth less than you paid for it.

Build, size and portability

What 3.116kg means

3.116kg is the number that should decide this purchase for a lot of people, and it deserves to be understood in concrete terms rather than as an abstract specification.

For reference points that do not require inventing figures: a 14in gaming laptop like the Razer Blade 14 in our rival list exists in an entirely different weight class, and a 16in machine sits somewhere in between. Three kilograms and change is heavier than most people expect a laptop to be, and the weight is only half of the problem — the other half is the footprint. An 18in display means a chassis with a footprint approaching that of a small desk. It will not fit comfortably on an aeroplane tray table, it will dominate a café table, and putting it in a backpack requires a backpack specifically bought for the purpose.

Then there is the charger. We do not have a measured figure for the power adapter's weight, so we will not state one, but a machine running a high-power CPU and a GPU near the top of its power range needs a substantial brick, and that brick travels with the laptop or the laptop does not run at full performance. Realistic transported mass is meaningfully above the 3.116kg figure.

The correct mental model is a desktop replacement. This is a machine that lives on a desk, moves between rooms occasionally, and travels between homes rather than between meetings. Used that way it is excellent — the large panel, the strong cooling that the benchmark figures imply, and the full-power GPU are all direct consequences of the size, and if the size is not a cost to you then it is pure benefit. Used as an actual portable computer it will be a persistent, daily irritation, and you will end up leaving it where it is and resenting having paid for a battery.

The battery

The 90Wh capacity is close to the practical maximum for a laptop — regulations effectively cap airline-carried lithium batteries around the 100Wh mark, so 90Wh is a big battery by any measure.

We do not have a measured runtime figure for this machine and we are not going to estimate one. What is worth saying is structural rather than numerical: 90Wh sounds generous but is being asked to supply an 18in high-resolution 240Hz panel and a Core Ultra 9 275HX. Panel size and refresh rate are among the largest contributors to idle and light-use power draw, and a high-core-count HX-series CPU has a higher floor than a lower-tier chip. The battery is large in absolute terms and modest relative to what it has to feed.

More importantly, gaming on battery on a machine like this is not a real proposition regardless of capacity. A GPU running near the top of the RTX 5070's power band cannot be supplied from a laptop battery at anything approaching full performance; every machine in this class throttles substantially when unplugged. Assume that all of the frame rates in this review require mains power, because they do.

What we do not know

In the interests of the site's own rules: we do not have measured dimensions, keyboard or trackpad assessments, port configurations, build quality observations, surface temperatures, fan noise levels or display calibration data for this machine. We are working from published specifications and published measurements, and where we do not have a figure we will say so rather than fill the gap. If those factors are decisive for you — and keyboard quality in particular is a legitimate reason to reject a laptop outright — you need hands-on impressions from someone who has had the machine in front of them.

How it compares

Cross-machine comparisons here are drawn only from games both machines have measured results for, which is why the number of shared titles varies. A comparison over eight or nine shared games is considerably more robust than one over five, and we have flagged the count in each case.

Alienware 16X Aurora — the one that beats it

The Alienware 16X Aurora runs the same RTX 5070 and costs £2,569. Across 6 shared games, the Crosshair averages -4fps against it. So the Alienware is very slightly faster and costs £474 more.

A 4fps average deficit across six titles is small — small enough that it could plausibly be accounted for by a modestly higher GPU power limit, a different cooling design, or run-to-run variance in a couple of the titles. In practical terms these two machines perform the same. You will not perceive the difference in any game.

What you will perceive is £474. That is a substantial sum, and it buys you nothing measurable in frame rate. This is the comparison that most flatters the Crosshair: it delivers essentially the same gaming performance as a machine costing nearly a quarter more, and it does so on a larger screen. If the choice is genuinely between these two on gaming performance alone, the Crosshair is the better buy by a wide margin — though the Alienware is a 16in machine, which means it is the more sensible option for anyone who needs to move it.

MSI Katana 17 HX — beaten on both counts

MSI's own Katana 17 HX, also with an RTX 5070, costs £2,312. Across 5 shared games, the Crosshair averages +7fps against it.

This one is straightforward: the Crosshair is faster and £217 cheaper. Five shared games is a thin sample and a 7fps average is not a large margin, but there is no direction in which the Katana wins here. It is a smaller machine — 17in rather than 18in — which is worth something if you care about footprint, but not £217 plus a performance deficit. Within MSI's own range, the Crosshair 18 HX AI is the better-judged product at these prices.

It is also worth noting what this comparison says about the market. Two RTX 5070 machines from the same manufacturer, £217 apart, with the cheaper one faster. Price at this tier is not tracking performance in any reliable way, which is exactly why buying on specification badges alone is a mistake and why frame-rate data matters.

Razer Blade 14 — the power limit made visible

The Razer Blade 14 runs the same RTX 5070 at a confirmed 115W. Across 8 shared games, the Crosshair averages +12fps against it. There is no confirmed UK price for the Blade 14 in our database, so we cannot make a value comparison.

Eight shared games is a solid sample, and 12fps is a real, perceptible gap. This is the clearest illustration available of why "RTX 5070" on a spec sheet is not a performance figure. The same GPU in a 14in chassis at 115W is meaningfully slower than the same GPU in an 18in chassis with room to breathe. If you are choosing between machines on GPU model, this comparison should change how you shop.

It also frames the trade-off honestly. The Blade 14 is a fundamentally different proposition — a machine you can actually carry, in a chassis a fraction of this one's size. The 12fps is the price of that portability. Whether that is a good trade depends entirely on whether you will ever put the machine in a bag. For a desk-bound buyer, giving up 12fps to get a smaller machine you never move is pure loss. For someone who commutes, it may be the whole point.

HP Omen 17z-db100 — close, but no confirmed price

The Omen 17z-db100 also runs an RTX 5070. Across 9 shared games — the largest sample in this comparison set — the Crosshair averages +4fps against it. We do not have a confirmed UK price.

Four frames across nine games is close to a tie. These two machines perform equivalently, and the Omen is a 17in design against this one's 18in. Without a UK price we cannot say which is the better buy, and that is a genuine gap rather than a hedge — if the Omen turns up at a meaningfully lower price, it becomes the obvious recommendation over the Crosshair for anyone who does not specifically need the larger panel. If you are shopping this segment, check the Omen 17z's current UK availability before committing.

HP Omen 16-ap0091ng — the tier below

This one runs an RTX 5060 rather than a 5070, and across 7 shared games the Crosshair averages +18fps against it. No confirmed UK price.

An 18fps average across seven games is the largest gap in this comparison set and it quantifies the step between GPU tiers usefully. It is a real difference — roughly the gap between a game running at 115fps and one running at 97fps, or between 62fps and 44fps at the demanding end, where it matters most. If you are weighing a 5060 machine against a 5070 machine, this is roughly the size of the prize.

Whether it is worth the price difference depends on a price we do not have. What we can say is that the gap is large enough to be worth paying something for, particularly given that the 5060 tier will hit the demanding-title wall sooner. But 18fps is not transformative either, and if the 5060 machine is several hundred pounds cheaper it is very likely the better value purchase.

The pattern across all five

Step back from the individual comparisons and a clear picture emerges. Against four other RTX 5070 machines, this one is faster than three and marginally slower than one. It is the cheapest of the two rivals with confirmed UK prices. Its performance sits at or near the top of what the RTX 5070 delivers in a laptop, and the Razer Blade 14 comparison suggests that advantage comes from running the GPU harder than a smaller chassis can.

So on a relative basis, this is a well-executed machine. The problem is that relative comparison within a badly-priced segment does not make the segment good. Every machine in this list is an RTX 5070 with 8GB of memory, and the two with confirmed prices are £2,095 and £2,569 and £2,312. That is a lot of money for this level of performance, and being the best of that group is a narrower achievement than it sounds. The Crosshair wins its comparisons and still costs more than the frame rates justify.

Who it's for

The case for buying it

You want the biggest screen available and you will actually use it. This is the strongest argument. 18in at 2560x1600 with a 240Hz refresh rate is close to the best non-OLED panel configuration in a laptop, and there are not many machines that offer it. If a large share of your computing happens on this screen — games, yes, but also work, media, multiple windows side by side — then the panel is doing real work every hour you own the machine, and paying a premium for it is rational in a way that paying a premium for frame rates you cannot perceive is not.

You need the CPU as much as the GPU. The Core Ultra 9 275HX is a serious processor, and the X-Plane 11 and Dota 2 results show it performing. If your workload includes compilation, encoding, simulation, heavily-modded games, or anything else where core count and single-thread speed dominate, a substantial part of the £2,095 is buying you something a cheaper machine with the same GPU would not.

The machine is not going to move. If it sits on a desk and gets carried to a different room twice a year, the weight and footprint cost you nothing and buy you thermal headroom, a full-power GPU, and the largest panel on the market. The Razer Blade 14 comparison — 12fps across eight games — is the size of what that chassis is earning you.

You play a lot of competitive and older titles. Strange Brigade at 280fps, F1 24 at 223fps, Dota 2 at 182fps and GTA V at 177fps mean the high-refresh panel is genuinely exercised in that category, and the CPU means it will stay that way.

You are cross-shopping the Alienware 16X Aurora. Same GPU, essentially the same measured performance across six shared games, £474 more. If those are your two candidates, this is the one.

The case against

You are buying on frames per pound. Then this is not the machine and, frankly, this is not the tier. £2,095 for a scoring-suite average of 141fps at 1080p is poor value against what the market offers, and the benchmark data above is the argument — 62fps in Black Myth: Wukong is the number that tells you what you are actually getting in a current-generation title. There is nothing wrong with the performance. There is something wrong with the price attached to it.

You want to play at the panel's native resolution without upscaling. Every measured figure here is at 1920x1080, and 2560x1600 is over twice the pixel count. In the demanding titles, native-resolution High-preset gaming with upscaling off is not what this GPU is for. If that is your requirement, you need a tier up, and you should budget accordingly rather than buying this and being disappointed.

You keep laptops for five or more years. The 8GB of video memory is the constraint that will bite first, it cannot be changed, and the panel's resolution makes it bite slightly harder. In four years the GPU will still be producing serviceable frame rates and you may be turning textures down to get them. That is the specific compromise that most visibly degrades how a game looks.

The machine needs to be portable. 3.116kg plus an adapter, in an 18in footprint, is a desktop replacement and nothing else. If you commute, travel, work from cafés, or take a laptop to a friend's house with any regularity, this will make you unhappy, and the Blade 14 comparison shows that the portability penalty is only 12fps across eight games. That is a very reasonable price to pay for a machine you can actually carry.

You want an OLED panel. This is IPS. It is a good, sensible, burn-in-proof choice, and it will not match OLED for black level or response time. If contrast in dark games is your priority, look elsewhere.

What would change the verdict

Price, almost entirely. Nothing about this machine's measured performance, panel or configuration is a problem — it is a well-built specification with the GPU running hard and a display that most of its rivals cannot match. What holds it back is that £2,095 buys an RTX 5070 with 8GB of memory, and that combination is not worth this money against what else the market offers. A meaningful discount would move it from "hard to recommend" to "straightforwardly good", and given how

How it compares