Razer Blade 16 Core Ultra 9 review

Frame rates

1920×1080, High preset, upscaling off. Averages 152fps across the 5 core benchmark games.

Measured on this machine. Source: Notebookcheck, 2026-05-27.

In use

The short version

The Razer Blade 16 with a Core Ultra 9 386H and an RTX 5090 is a very fast laptop that costs £5,000, and the second of those facts is the one that decides the review. Notebookcheck's numbers put it comfortably in the top tier of what mobile silicon can do: an average of 152fps across the five games in our scoring suite, at 1920x1080 on High with upscaling switched off, with nothing in the ten measured titles dropping below 86fps. On raw output there is very little to complain about.

The problem is everything around that number. A Medion Erazer Beast 16 X1 Ultimate with the same RTX 5090 costs £4,884 — £116 less — and averages 22fps faster than the Blade across the six games both machines have been measured on. An Asus ROG Strix Scar 16, also RTX 5090, averages 24fps faster across eight shared games, though we have no confirmed UK price for it and so cannot say what that speed costs. Further down, an Alienware 16X Aurora with an RTX 5070 at £2,569 gives up an average of only 8fps across seven shared titles, and a Lenovo Legion 5 Pro with an RTX 5070 Ti is level with the Blade — an average difference of zero across eight shared games. That last figure deserves reading twice. On the games both machines have been measured on, the £5,000 RTX 5090 Blade and a mid-range RTX 5070 Ti Legion produce, on average, the same frame rate.

So what is the £5,000 buying? A 16in 2560x1600 OLED panel at 240Hz, which is as good a specification as a gaming laptop screen currently gets; a 2.124kg chassis, which is light for a 16in machine carrying this class of GPU; 32GB of memory, 2TB of storage and 24GB of VRAM, none of which you will run out of soon. Those are real things and they are worth real money. They are not worth the roughly £2,400 gap between this and the Alienware, and they are certainly not worth paying more than a machine with the same GPU that runs faster.

Buy it if the combination of that panel, that weight and that GPU in one chassis is genuinely the thing you need, and the price is not the binding constraint — if you are carrying it daily, working on it, and the OLED is doing colour work as well as gameplay. Everyone else is being asked to pay a very large premium for a specific set of physical qualities rather than for frames. That is poor value at this price, and the comparison figures below show exactly how poor.

Performance

All the frame rates here come from Notebookcheck, measured on 2026-05-27 at 1920x1080, High preset, with upscaling off. That last condition matters more than people usually credit: these are native-render numbers with no DLSS, no frame generation and no reconstruction of any kind. They represent what the silicon actually draws, which makes them a clean basis for comparison but also means they are not the numbers you will see on screen if you turn DLSS on, which on this GPU you almost certainly will in the heavier titles.

The measured spread

Here is the full set, ordered:

• Strange Brigade — 353fps
• F1 24 — 258fps
• Baldur's Gate 3 — 196fps (scoring suite)
• Final Fantasy XV — 176fps
• Dota 2 — 162fps (scoring suite)
• GTA V — 156fps (scoring suite)
• Cyberpunk 2077 — 149fps (scoring suite)
• X-Plane 11 — 98fps (scoring suite)
• Black Myth: Wukong — 97fps
• Assassin's Creed Shadows — 86fps

Top to bottom that is a spread of 353fps down to 86fps — a factor of just over four. That range is normal for a benchmark suite that deliberately mixes eras and engines, but the shape of it tells you a lot about where this machine is and is not limited.

The top of the list is not about the GPU

Strange Brigade at 353fps and F1 24 at 258fps are the two titles clearing the panel's 240Hz refresh at 1080p, and neither is a meaningful test of an RTX 5090. Strange Brigade is a well-optimised, Vulkan-friendly benchmark that has been used for years precisely because it scales cleanly and runs fast; at 353fps the bottleneck is not the graphics card in any interesting sense. F1 24 at 258fps is a similar story — a racing title with a tightly controlled draw call budget and a rendering path that has been polished over many annual releases.

What these two numbers are genuinely useful for is establishing the platform's ceiling. When a machine can push 353fps in a title that will take everything you give it, you know the CPU, memory subsystem and driver stack are not getting in the way at the top end. The Core Ultra 9 386H and 32GB of memory are feeding the GPU properly. That sounds like faint praise but it is not: plenty of laptops with excellent GPUs post disappointing numbers in the easy titles because something upstream is choking, and that same choke then quietly costs frames in the harder games too.

The middle band is the honest picture

Baldur's Gate 3 at 196fps, Final Fantasy XV at 176fps, Dota 2 at 162fps, GTA V at 156fps and Cyberpunk 2077 at 149fps form a tight cluster between roughly 150 and 200fps, and this band is where the machine's real character sits. Four of those five are in our scoring suite, which is why the suite mean lands at 152fps.

Cyberpunk 2077 at 149fps on High at 1080p with no upscaling is a strong result in absolute terms. Cyberpunk remains, six years on, a reliable stress test of a rasterisation pipeline — dense geometry, heavy volumetrics, expensive screen-space effects — and pushing it to just short of 150fps natively is something only the top mobile GPUs manage. GTA V at 156fps sitting next to Cyberpunk at 149fps is more interesting than it looks. GTA V is a game from 2013 that should, on hardware of this class, be running vastly faster than a 2020 open-world title. That it is only 7fps ahead is a strong signal that GTA V here is limited by something other than the GPU — its ageing engine's CPU-side draw submission, most likely — and it is a good reminder that a chunk of the scoring suite mean is being set by factors the RTX 5090 cannot influence.

The same applies, more sharply, to Dota 2 at 162fps. Dota 2 is a game a decade and a half old with modest per-pixel cost, and on a machine capable of 353fps in Strange Brigade you would expect it substantially higher than 162fps unless the test scene is deliberately punishing or the game is hitting a CPU or engine ceiling. Notebookcheck's Dota 2 run is a demanding one, and whatever the precise cause, the practical implication is the same: this is not a GPU-limited result, and the RTX 5090 is not the component that would fix it. If you play a lot of Source-engine or similarly CPU-bound competitive titles, the graphics card in this laptop is not the part that determines your experience, and you are paying an enormous premium for a component that is idling.

The bottom of the list is where the GPU earns its keep — and where the power limit shows

X-Plane 11 at 98fps, Black Myth: Wukong at 97fps and Assassin's Creed Shadows at 86fps are the three genuinely demanding results, and they are the ones worth thinking hardest about.

Black Myth: Wukong at 97fps is an Unreal Engine 5 title with Lumen-class lighting and Nanite geometry, running natively at 1080p on High. Ninety-seven frames per second there is a good number — it is the kind of result that separates a top-tier mobile GPU from a mid-tier one, because UE5 workloads scale much more directly with raw shader throughput and memory bandwidth than the older engines higher up the list do. Assassin's Creed Shadows at 86fps is the lowest figure in the set and, again, a genuinely heavy modern open-world renderer with expensive global illumination and long draw distances.

X-Plane 11 at 98fps is a different animal. Flight simulators are notoriously CPU- and draw-call-bound, and X-Plane in particular has historically been more sensitive to single-thread performance and scenery streaming than to shader power. Its presence in the scoring suite pulls the mean down, but it is not really telling you about the RTX 5090; it is telling you about the platform as a whole under a specific and unusual kind of load. If you fly, that number is directly relevant to you. If you do not, treat it as a data point about the machine's breadth rather than about its graphics capability.

What the GPU power limit means here

This is the part where we have to be straight about what we do not know. Our fact sheet does not include a confirmed graphics power limit for this configuration, and we are not going to guess at a wattage figure. But the benchmark data lets us reason about it indirectly, and the reasoning is unflattering.

The RTX 5090 in a laptop is not a single fixed product. It is a silicon design that manufacturers configure within a wide band of sustained power, and the difference between a chassis that runs it at the top of that band and one that runs it near the bottom is not marginal — it is routinely the difference between two tiers of GPU. Two laptops can both be advertised with an RTX 5090 and produce meaningfully different frame rates in the same game at the same settings, because one has the thermal headroom and power delivery to hold the clocks and the other does not.

Now look at the comparison data. Against the Medion Erazer Beast 16 X1 Ultimate — also an RTX 5090 — the Blade averages 22fps slower across six shared games. Against the Asus ROG Strix Scar 16 — also an RTX 5090 — it averages 24fps slower across eight shared games. Same GPU, same benchmark suite, same methodology, consistently behind. There is no plausible explanation for a gap of that size and consistency between machines with identical graphics silicon other than power and thermal configuration. The Blade is not extracting as much from its RTX 5090 as thicker, heavier rivals are.

That is not a scandal — it is the deal you make when you buy a 2.124kg 16in laptop rather than a 2.7kg one. A thinner chassis has less volume for heatsink mass, less area for fin stack, less room for airflow, and a lower sustainable power envelope as a direct consequence. Razer's engineering trade-off is deliberate and it is the entire point of the Blade line. But it does change what the badge on the box means. You are buying an RTX 5090 that behaves, in sustained load, somewhere between where the fastest RTX 5090 laptops sit and where a strong RTX 5070 Ti machine sits — and the Legion 5 Pro comparison, level on average across eight shared games, is the clearest possible evidence of that.

The practical consequence for a buyer is this: the frame rates that matter most to you are the ones at the bottom of the list, in Wukong and Shadows and anything else built on a modern engine, and those are precisely the titles where a lower sustained power limit costs the most. In Strange Brigade the power limit is irrelevant because nothing is asking the GPU to work hard. In Assassin's Creed Shadows it is the whole ballgame. The 86fps figure is respectable; it would very likely be higher in a chassis with more thermal headroom, and the rival data strongly suggests it is.

The scoring-suite mean, and why it is the right number to hold on to

The mean across the five scoring-suite games — Baldur's Gate 3, Dota 2, GTA V, Cyberpunk 2077 and X-Plane 11 — is 152fps. That is the figure we weight performance on, and it is worth understanding what it does and does not capture.

It excludes Strange Brigade and F1 24, which is the right call: including a 353fps result in an average would let one easy title distort the whole picture, and a machine that is brilliant at Strange Brigade and mediocre at Cyberpunk should not be flattered by arithmetic. It includes X-Plane 11 at 98fps, which drags the mean down but reflects a real category of workload. And it spans a genuine range within itself — 98fps to 196fps — which means the mean sits roughly in the middle of a wide distribution rather than summarising a tight cluster.

What 152fps means in practice: on this machine, at 1080p, with no upscaling, on High, you are comfortably above the 120fps mark in most of what you will play, and above 60fps in everything measured. That is a strong performance profile. It is not a profile that is unique to £5,000 hardware, which is the problem.

What the data does not tell us

Three gaps are worth naming explicitly, because reviews that quietly paper over them are doing you a disservice.

All the measured frame rates are at 1920x1080, and the panel is 2560x1600. That is 4.1 megapixels versus 2.1 — very close to double the pixel count. We do not have measured native-resolution figures for this machine, so we are not going to state what it does at 2560x1600. What we can say honestly is that in the GPU-bound titles at the bottom of the list, doubling the pixel count is a substantial cost, and in the CPU-bound titles at the top it is a much smaller one. Anyone reading 149fps in Cyberpunk and assuming that is what they will see on the built-in display at native resolution is going to be disappointed. Anyone reading 86fps in Assassin's Creed Shadows and assuming the same should be planning around DLSS.

There are no ray-tracing figures in this data set. Cyberpunk 2077 in particular has path-tracing modes that are among the heaviest workloads in consumer software, and the 149fps figure is a rasterisation result on High. The RTX 5090's ray-tracing hardware is one of the main reasons to buy it over a lesser card, and we cannot quantify what it does here. That is a genuine hole in the picture for a machine at this price.

We have no measured battery life, temperature or noise figures. The battery is 90Wh, which is at the practical upper end for a laptop, but a capacity figure is not a runtime figure, and a machine with a 90Wh pack and an RTX 5090 can post wildly different runtimes depending on how it handles graphics switching and display power. We are not going to invent a number. If battery life under load is central to your decision, this review cannot answer that question and you should look for measured runtime data before committing £5,000.

CPU, memory and storage

The Core Ultra 9 386H is doing its job. The evidence for that is the 353fps in Strange Brigade and the 258fps in F1 24 — a platform that could not keep up would not produce those. It is a mobile part in a thin chassis, and it shares a thermal budget with the GPU, which is worth remembering if your workload is heavily CPU-bound: in a thin machine, sustained all-core load and sustained GPU load are competing for the same headroom, and something has to give. The Dota 2, GTA V and X-Plane 11 figures are the ones most likely to be shaped by that dynamic.

32GB of memory is the right amount in 2026 — enough that no current game will touch the limit and enough for serious content work alongside. 2TB of storage is generous and, at this price, should be considered the minimum acceptable; a 1TB drive on a £5,000 laptop would have been an insult. 24GB of VRAM is more than any current game needs at 1080p or 1600p and is really there for the workstation and local-AI use cases, which is a reasonable justification for the RTX 5090 badge on a machine that will not fully exploit it in games. If you are running large models locally or working with high-resolution assets, that 24GB buffer is a genuine differentiator over the RTX 5070 Ti and RTX 5070 machines it otherwise resembles on frame rates. If you are only gaming, it is capacity you are paying for and will not use.

The screen

This is the part of the machine that needs no defending. A 16in OLED panel at 2560x1600 and 240Hz is, on paper, about as complete a specification as a laptop display currently offers, and it is the single clearest justification for the price gap between this and the £2,569 Alienware.

The resolution is right for the size. At 16in, 2560x1600 gives you a pixel density that makes text and UI genuinely crisp without being so high that you are throwing GPU cycles at pixels you cannot resolve — which is the trap of 4K panels at this diagonal. The 16:10 aspect ratio gives you vertical space that matters more than people expect once they have used it: more visible lines in a code editor, more timeline in an NLE, more of the world above and below the horizon in a game. If this machine is doing double duty as a work laptop, and at this price it probably should be, the aspect ratio is not a trivial detail.

OLED brings per-pixel illumination, which means true black and effectively infinite contrast, and it brings response times that no LCD can match. For a gaming display that second point is arguably the more important one: a 240Hz panel with slow pixel transitions delivers far less than 240Hz of actual motion clarity, and OLED simply does not have that failure mode. Combined with the refresh rate, this is a display that will show you every frame the GPU produces, cleanly.

The trade-offs of OLED are the ones you already know and our data cannot quantify: static-element burn-in risk over a long ownership period, and peak sustained full-screen brightness that typically trails a good mini-LED backlight. We have no measured brightness figure for this panel, so we are not going to characterise it. If you work outdoors or in a very bright room, find measured nit figures before you buy.

Can the GPU actually drive it?

This is the question that matters, and the honest answer is: partly, and it depends heavily on the game.

At the panel's native 2560x1600, we have no measured frame rates at all, so anything specific would be invention. What we can do is reason from the 1080p data about which titles have headroom and which do not.

Of the ten measured games, two exceed 240Hz at 1080p: Strange Brigade at 353fps and F1 24 at 258fps. Both of those are, as discussed, light on the GPU, which means both are the titles most likely to retain a large fraction of their frame rate when the resolution goes up — CPU-bound results move less with pixel count than GPU-bound ones. So the two games that saturate the refresh rate at 1080p are also the two with the best chance of staying near it at native. That is a happy coincidence for anyone whose main use is competitive racing.

Everything else is already below 240fps at 1080p, and will fall further at 4.1 megapixels. Baldur's Gate 3 at 196fps and Final Fantasy XV at 176fps have room to remain very playable. Cyberpunk at 149fps and GTA V at 156fps are in the band where native 1600p on High, without upscaling, is likely to land somewhere well short of the refresh ceiling but still comfortably smooth. Black Myth: Wukong at 97fps and Assassin's Creed Shadows at 86fps are the ones where native 1600p without reconstruction becomes a real question, and where DLSS stops being optional and starts being the intended way to play. That is not a criticism specific to this laptop — it is true of every mobile GPU in 2026 — but it does mean that the 240Hz refresh rate is a specification you will exploit in older and lighter titles and largely bank as motion clarity rather than raw frame delivery in the heaviest ones.

The other thing worth saying: 240Hz on an OLED is valuable even when you are not hitting 240fps. Higher refresh reduces the display's contribution to input latency and makes variable-refresh operation smoother across a wider range. A game running at 120fps on a 240Hz OLED looks and feels better than the same game at 120fps on a 120Hz LCD. So the panel is not wasted just because the GPU cannot always feed it.

Build, size and portability

What we have confirmed is a 16in machine weighing 2.124kg with a 90Wh battery. Our fact sheet does not give thickness or footprint dimensions, and we are not going to estimate them. But the weight figure alone carries a lot of information, and it is the second-strongest argument in this laptop's favour after the screen.

Two point one kilograms for a 16in chassis containing an RTX 5090 and a Core Ultra 9 is genuinely light for the category. The reference points that matter here are not ultrabooks — nothing with this GPU is going to compete with a 1.3kg thin-and-light — but other 16in and 17in gaming machines, which routinely land closer to 2.5kg and often well beyond it once you get into the flagship-GPU tier with the cooling to match. A 400g saving over a typical rival sounds abstract on a spec sheet and is entirely concrete in a rucksack on a commute.

The corollary, though, is the one already established in the performance section: the weight saving is not free, and the benchmark data shows where the bill lands. The Medion and the Asus, both carrying the same RTX 5090, are 22fps and 24fps faster on average across their shared games with this machine. Physics is not negotiable — heat has to go somewhere, and a chassis with less mass and less volume sheds it more slowly and sustains less power as a result. The Blade's proposition is that you would rather carry the lighter machine and give up those frames. For some buyers that is straightforwardly the correct trade. For others it is paying flagship money for sub-flagship sustained performance, which is a harder case to make.

The 90Wh battery is at the top of what is practical — regulations effectively cap carry-on lithium capacity around the 100Wh mark, so there is very little upside left to chase. Whether that translates into useful unplugged runtime we cannot say, because we have no measured battery figures. What is safe to say is that no laptop with an RTX 5090 runs games on battery for long, and that any realistic portable use of this machine means either light workloads unplugged or carrying a charger. A 16in gaming laptop's power adapter is not a small object and its weight is not included in the 2.124kg figure. If your definition of portable is "fits in a bag and works all day without mains power", no machine in this class satisfies it, this one included.

Where the weight genuinely pays off is the specific pattern of use this laptop is designed around: a machine that lives on a desk, gets carried between desks — home, office, studio, client — and is used plugged in at both ends. For that, 400g matters, and matters daily. For a machine that sits in one place permanently, it is worth nothing at all, and you would be better served by a heavier laptop with the same GPU running harder.

How it compares

A note on method before the numbers. Our comparison figures are averages of per-game frame rate differences, computed only across the titles both machines have been measured on. That is the fairest way to compare two devices with partially overlapping benchmark sets, but it has a known weakness: an average of raw fps differences is dominated by high-frame-rate titles. A 40fps gap in Strange Brigade at 350fps is practically meaningless — you will not perceive it, and the panel cannot show it — while a 15fps gap in Assassin's Creed Shadows at 86fps is the difference between smooth and not. So read these averages as directional evidence about where machines sit relative to one another, not as a precise statement of how much better one will feel. Where an average is computed over a small number of shared games, treat it more cautiously still.

Lenovo Legion 5 Pro (RTX 5070 Ti, no confirmed UK price)

Across eight shared games — a large and therefore reasonably reliable overlap — the Blade averages +0fps against the Legion 5 Pro. Level. Identical, on average, to a machine with a graphics card two tiers down.

This is the single most damaging comparison in the set, and it deserves to be understood properly rather than waved away. It does not mean an RTX 5070 Ti is as fast as an RTX 5090; it means this implementation of an RTX 5090, in a 2.124kg chassis with the sustained power that implies, produces the same average output across eight games as a well-cooled RTX 5070 Ti. Some of that will be the CPU-bound titles at the top of the suite, where both machines hit the same platform ceiling and the GPU difference cannot express itself. But eight games is a broad enough sample that the heavy titles are in there too, and the average still comes out flat.

We have no confirmed UK price for the Legion 5 Pro, so we cannot state the value gap in pounds. What we can say is that an RTX 5070 Ti laptop from Lenovo's mainstream gaming line is not a £5,000 product, and the Blade is not delivering more frames than it. Whatever the Legion costs, the Blade's premium is buying the OLED panel, the weight, the storage and the VRAM — not speed.

Asus ROG Strix Scar 16 (RTX 5090, no confirmed UK price)

Across eight shared games, the Blade averages 24fps slower. Same GPU, same benchmark methodology, and the largest deficit in our comparison set, measured over the largest overlap.

This is the clearest available evidence of what the thin chassis costs. The Strix Scar 16 is a bigger, heavier, more aggressively cooled machine built with sustained performance as its first priority, and the numbers reflect it. Twenty-four frames per second averaged over eight games is not noise and it is not measurement variance; it is a systematic difference in how much power each machine lets its RTX 5090 draw over the length of a benchmark run.

Without a confirmed UK price we cannot tell you whether the Scar represents better value, and we are not going to guess. But if raw frame rate from an RTX 5090 is your objective, the Blade is not the machine delivering it, and you should be pricing the Asus before you spend £5,000 here.

Medion Erazer Beast 16 X1 Ultimate (RTX 5090, £4,884)

Across six shared games, the Blade averages 22fps slower — and the Medion costs £116 less.

This is the comparison that makes the value case unarguable, because both variables point the same way. Same GPU class, cheaper, faster. Six shared games is a smaller overlap than the Asus comparison and so a slightly weaker instrument, but it corroborates the Asus result almost exactly — two independent RTX 5090 machines, both roughly 20-25fps ahead of this one on average. Two data points telling the same story is considerably more convincing than one.

The Medion is not the same kind of product. It is a big, functional, performance-first machine, and we have no weight or panel data for it in this comparison to set against the Blade's OLED and 2.124kg. If those two things matter enormously to you, the Blade still has an argument. But it is now an argument entirely about industrial design and display quality, made at a price premium over a faster machine, and it needs to be a very strong preference to justify.

Alienware 16X Aurora, RTX (RTX 5070, £2,569)

Across seven shared games, the Blade averages 8fps faster. It costs £2,431 more.

Eight frames per second. Averaged across seven games. For roughly £2,400. Whatever else is true about this review, that sentence is the one to keep.

Some caveats in fairness. Eight frames per second averaged across a suite that includes CPU-bound titles at 250fps and GPU-bound titles at 86fps is not evenly distributed — the gap in Assassin's Creed Shadows or Black Myth: Wukong will be considerably larger than eight frames, and the gap in Strange Brigade or Dota 2 close to nothing. The Blade's advantage is concentrated exactly where it counts, in the heavy modern engines. And the Alienware is an RTX 5070 with a smaller VRAM allocation, which will matter more as time passes and asset budgets grow. The Blade is more future-proofed.

But £2,431 is not a rounding error. It is another entire gaming laptop. It is the difference between the Alienware and nothing, twice over. If the seven-game average gap is eight frames, the Blade has to justify that money on the panel, the weight, the storage, the memory and the VRAM — and while all of those are better, they are not £2,431 better for most buyers.

MSI Katana 17 HX (RTX 5070, £2,312)

Across five shared games, the Blade averages 18fps faster, for a price difference of £2,688.

Five shared games is the thinnest overlap in the set and the average should be treated with corresponding caution — one heavy title can move a five-game average substantially. Taken at face value, though, the Katana is the crudest possible expression of the value question: a 17in machine, a lesser GPU, and less than half the price, giving up an average of 18fps.

The Katana is a fundamentally different proposition — bigger, cheaper, built to a budget, and by any reasonable expectation heavier and with a lesser display, though we have no confirmed panel or weight data for it here to state that. Nobody cross-shopping a Blade 16 is seriously considering a Katana 17 HX. But it establishes the floor: this is roughly what an RTX 5070 gaming laptop costs and roughly what it delivers, and everything above it is priced against that baseline. The Blade sits at more than twice the cost for an average gain that, in the games we can compare, is meaningful but not transformative.

What the comparison set says as a whole

Line the five up and a clear pattern emerges. The Blade is slower than both other RTX 5090 machines, by 22fps and 24fps over six and eight shared games respectively. It is level with an RTX 5070 Ti machine over eight shared games. It is modestly ahead of two RTX 5070 machines, by 8fps and 18fps, both of which cost less than half as much.

The performance is genuinely good in absolute terms — 152fps across the scoring suite is a top-tier result and nothing measured falls below 86fps. But this machine's position relative to its actual competition is upper-middle, not top, despite carrying the top GPU and the top price. The RTX 5090 badge is not delivering RTX 5090 output in this chassis, and the two rivals with the same silicon prove it.

Who it's for

There is a real buyer for this laptop and it is worth describing them precisely, because generic "it's expensive but premium" hedging helps nobody spending £5,000.

Buy it if you carry it every day and the display is doing real work. The combination of 2.124kg, a 16in 2560x1600 OLED and an RTX 5090 in one machine is not common, and if you are a photographer, colourist, motion designer or developer who also games seriously, and who moves between locations several times a week, this is a coherent product. The OLED's contrast and colour behaviour is a professional tool, the 16:10 ratio gives you working vertical space, the 24GB of VRAM handles asset-heavy or local-model workloads that would choke a 12GB card, and the weight means you will actually take it with you. That buyer is getting value the frame rates do not capture.

Buy it if you specifically want the thin-flagship form factor and understand the trade. Some people simply do not want a 2.7kg brick with an aggressive aesthetic, and are willing to pay for restraint and a small footprint. That is a legitimate preference and this is one of the few machines that serves it at this GPU tier. Just go in knowing that the Medion and the Asus, on this data, are faster with the same graphics card, and that the difference is not small.

Buy it if your games are the ones at the bottom of the list. If you mostly play current-generation Unreal Engine 5 titles and heavy open-world games — the Black Myth: Wukong and Assassin's Creed Shadows end of the suite — the gap between this machine and the RTX 5070 laptops will be larger than the seven- and five-game averages suggest, because those averages are diluted by titles where nothing separates them. The 97fps and 86fps figures are the ones that make the GPU worth having.

Who should buy something else

If you want the fastest RTX 5090 laptop, this is not it. The Medion Erazer Beast 16 X1 Ultimate is £116 cheaper and averages 22fps faster across six shared games. The Asus ROG Strix Scar 16 averages 24fps faster across eight. If frames per pound from a flagship GPU is the objective, the Medion has the Blade beaten on both axes simultaneously and there is no argument to be had.

If the laptop is going to sit on a desk permanently, buy something heavier. The entire premium here is for portability and industrial design. A machine that never moves gains nothing from 2.124kg and loses real frame rate to the cooling constraints that weight implies. Buy a chassis with room to breathe and spend the difference on a monitor.

If you mostly play competitive or older titles, the GPU is wasted. Dota 2 at 162fps and GTA V at 156fps sit right next to Cyberpunk at 149fps, which tells you those results are set by the platform rather than the graphics card. An RTX 5070 machine will produce very similar numbers in that category of game for a fraction of the money. The Alienware 16X Aurora at £2,569 gives up an average of 8fps across seven shared games and would be indistinguishable in the titles you actually play.

If £5,000 is a stretch, it is emphatically the wrong machine. The Alienware at £2,569 and the MSI Katana 17 HX at £2,312 both deliver a large fraction of this performance for well under half the outlay. The Legion 5 Pro with its RTX 5070 Ti matches the Blade's average across eight shared games outright, and while we cannot confirm its UK price, it is not a £5,000 laptop. Spending an extra £2,400 to gain 8fps on average is not a decision that survives contact with a spreadsheet.

If you need measured battery, thermal or acoustic data before committing, we cannot give it to you. Our fact sheet has the 90Wh capacity and nothing about runtime, temperature or noise. On a £5,000 purchase where portability is the headline feature, that is a gap worth filling from another source before you buy.

The bottom line

The Razer Blade 16 with a Core Ultra 9 386H and RTX 5090 is a fast, beautifully specified laptop with one of the best displays available and a weight that genuinely distinguishes it. Nothing in the measured data is disappointing in absolute terms: 152fps across the scoring suite, 149fps in Cyberpunk 2077, 97fps in Black Myth: Wukong, 86fps in Assassin's Creed Shadows, all native at 1080p on High with no upscaling.

But it is beaten on frame rate by two other RTX 5090 machines, one of which costs less, and matched on average by an RTX 5070 Ti machine across eight shared games. At £5,000, against an £2,569 Alienware that trails by an average of 8fps and a £4,884 Medion that leads by 22fps, the value case is not weak — it is absent. This is a laptop you buy for its panel, its weight and its build, having decided that those are worth several thousand pounds to you. If they are, it is an excellent machine. If they are not, the benchmark data above says plainly where your money goes further.

How it compares