If there were ever a laptop design to be right "up my alley", as we brits like to say, it would be this one. That's because it makes its sacrifices in exactly the right places to deliver something genuinely portable, sleek, elegant, and satisfying to use both for gaming and work.

The MSI 'Stealth' laptop looks to have been reimagined along these MacBook or Razer Blade-esque lines—and successfully so—featuring a giant touchpad, thin aluminium form factor, subtle upward-facing speakers, and a spacious screen.

It's not just a looker, though, as MSI has crammed an RTX 5080 mobile GPU into it, making for genuine high-end performance. It has, however, had to be power-limited to play nicely with the slim chassis, meaning you're not getting quite the performance you could get out of a chunkier laptop for the same very expensive price or cheaper.

MSI Stealth 16 AI+ B3W Features

After powering on the laptop for the first time, I was immediately impressed by how it feels to use. The trackpad doesn't require much force to press but just enough so you're not accidentally pressing it, and it has a very satisfying click. I didn't have many problems with accidentally moving the cursor while typing because of the size of the trackpad, but it did happen occasionally.

'Satisfying' is also the word I'd use to describe the pleasantly tactile chiclet keyboard. This keyboard has four-zone RGB that I set to solid white to match the classy look.

There's also a decent selection of ports on the Stealth, including two Thunderbolt 4, a HDMI, and even an Ethernet port. This, I found to be particularly impressive given the sleek form factor.

Sleek, that is, and very shiny thanks to its 240 Hz OLED. It's of the 16:10 variety, too, which is the best kind as it gives you plenty of space for browsing and so on. I'd say 16-inch with a 16:10 aspect ratio is pretty much ideal for a laptop, especially one like this that's also slim and portable. The OLED panel itself is fantastic, too, with plenty of contrast thanks to the true blacks and vibrant colours. And of course the smooth 240 Hz certainly doesn't hurt in those games that can push frame rates up that high.

Specs-wise, this particular version gives you only the choice of the Core Ultra 9 386H for the processor, but you can choose between an RTX 5070, 5070 Ti, or 5080 version. The latter two come with 32 GB of DDR5 RAM, but the RTX 5070 version comes with just 16 GB. All configs except for one 2 TB version comes with 1 TB of storage.

The version I tested has 1 TB of storage, 32 GB of DDR5-5600 memory, and an RTX 5080 mobile. The RTX 5080 here has a big asterisk: it's a nerfed—ie, power-limited—version of the GPU. With a max draw of just 125 W, that's 50 W less juice than you get with a full-fat RTX 5080 mobile like the one in the Lenovo Legion 7i Pro Gen10, the current best high-performance gaming laptop. That's a sacrifice made to keep things so slim, as is the choice of CPU which sacrifices some top-end compute performance for efficiency.

So far so good, but there is one area that unfortunately rains all over the parade, and that's audio. The problem isn't the sound quality or volume, as things are great in both these respects—clear, punchy (though not as bassy as standalone speakers, of course), and plenty loud.

The problem is latency and glitching, though mainly the former. Initially the speakers would spit every now and then, but that seemed to clear up after using the laptop for a while. I'm not sure why that is, given I had Windows update paused, but whatever the reason, I was glad when it stopped.

The persistent problem was poor audio latency. There is a delay between my in-game action (such as shooting) and the sound coming through. It's about the same level of delay as I experience using Bluetooth earphones without any low-latency mode enabled. It only happens over the default audio device, not when I plug in a new USB device like my earphones with USB-C DAC.

That leads me to believe it might be a problem with the default audio device drivers, and after checking various (x) discussions (x) online it seems like the problem could be with Nahimic, however I cannot confirm this. I can say that I tried different methods to disable and uninstall the drivers, which didn't work, and I also uninstalled the Norton antivirus that comes bundled with the laptop just to see if that was the issue, but that didn't resolve it. MSI says it is looking into the issue, but so far I haven't heard any results of this investigation.

To be clear, using the speakers is fine for listening to music and for playing casual, non-shooter games, but for playing shooters the latency is definitely noticeable, meaning a separate USB audio device might be necessary. Though saying that, when testing a different MSI laptop, even switching to a different device didn't resolve the latency, though it strangely did with the Stealth.

MSI Stealth 16 AI+ B3W Performance

Given the high-end components sitting inside this MSI Stealth 16 AI+ B3W, it was never going to fail at churning out frames. You'd have to severely throttle an RTX 5080 laptop GPU for it to be unable to deliver playable performance. But for the Stealth's price tag we want more than just 'playable', so let's see how it does.

1080p gaming performance

This chart shows average and 1% low frame rate results in various tested games at 1080p resolution

Cyberpunk 2077 (1080p RT Ultra)Baldur's Gate 3 (1080p Ultra)Black Myth: Wukong benchmarking tool (1080p High)F1 24 (1080p Ultra High)Metro Exodus Enhanced Edition (1080p Ultra)

Avg FPS

1% Low FPS

MSI Stealth 16 AI+ | RTX 5080 (125 W)

55

32

Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)

52

39

Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)

68

38

Framework 16 | RTX 5070

43

33

020406080

Cyberpunk 2077 (1080p RT Ultra) Data

Product	Value
MSI Stealth 16 AI+ | RTX 5080 (125 W)	55 Avg FPS, 32 1% Low FPS
Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)	52 Avg FPS, 39 1% Low FPS
Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)	68 Avg FPS, 38 1% Low FPS
Framework 16 | RTX 5070	43 Avg FPS, 33 1% Low FPS

Avg FPS

1% Low FPS

MSI Stealth 16 AI+ | RTX 5080 (125 W)

68

39

Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)

78

43

Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)

96

55

Framework 16 | RTX 5070

69

41

0255075100

Baldur's Gate 3 (1080p Ultra) Data

Product	Value
MSI Stealth 16 AI+ | RTX 5080 (125 W)	68 Avg FPS, 39 1% Low FPS
Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)	78 Avg FPS, 43 1% Low FPS
Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)	96 Avg FPS, 55 1% Low FPS
Framework 16 | RTX 5070	69 Avg FPS, 41 1% Low FPS

Avg FPS

1% Low FPS

MSI Stealth 16 AI+ | RTX 5080 (125 W)

73

47

Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)

102

83

Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)

99

84

Framework 16 | RTX 5070

64

55

037.575112.5150

Black Myth: Wukong benchmarking tool (1080p High) Data

Product	Value
MSI Stealth 16 AI+ | RTX 5080 (125 W)	73 Avg FPS, 47 1% Low FPS
Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)	102 Avg FPS, 83 1% Low FPS
Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)	99 Avg FPS, 84 1% Low FPS
Framework 16 | RTX 5070	64 Avg FPS, 55 1% Low FPS

Avg FPS

1% Low FPS

MSI Stealth 16 AI+ | RTX 5080 (125 W)

100

50

Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)

101

76

Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)

126

51

Framework 16 | RTX 5070

75

58

037.575112.5150

F1 24 (1080p Ultra High) Data

Product	Value
MSI Stealth 16 AI+ | RTX 5080 (125 W)	100 Avg FPS, 50 1% Low FPS
Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)	101 Avg FPS, 76 1% Low FPS
Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)	126 Avg FPS, 51 1% Low FPS
Framework 16 | RTX 5070	75 Avg FPS, 58 1% Low FPS

Avg FPS

1% Low FPS

MSI Stealth 16 AI+ | RTX 5080 (125 W)

102

42

Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)

102

53

Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)

140

72

Framework 16 | RTX 5070

82

51

037.575112.5150

Metro Exodus Enhanced Edition (1080p Ultra) Data

Product	Value
MSI Stealth 16 AI+ | RTX 5080 (125 W)	102 Avg FPS, 42 1% Low FPS
Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)	102 Avg FPS, 53 1% Low FPS
Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)	140 Avg FPS, 72 1% Low FPS
Framework 16 | RTX 5070	82 Avg FPS, 51 1% Low FPS

The MSI Stealth 16 AI+ sits between RTX 5070 and full-fat (175 W) RTX 5080 gaming laptops in 1080p performance.

1600p (native) gaming performance

This chart shows average and 1% low frame rate results in various tested games at 1600p (native) resolution

Cyberpunk 2077 (Native res. RT Ultra)Baldur's Gate 3 (Native res. Ultra)Black Myth: Wukong benchmarking tool (Native res. High)F1 24 (Native res. Ultra High)Metro Exodus Enhanced Edition (Native res. Ultra)

Avg FPS

1% Low FPS

MSI Stealth 16 AI+ | RTX 5080 (125 W)

34

22

Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)

31

24

Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)

40

28

Framework 16 | RTX 5070

18

8

010203040

Cyberpunk 2077 (Native res. RT Ultra) Data

Product	Value
MSI Stealth 16 AI+ | RTX 5080 (125 W)	34 Avg FPS, 22 1% Low FPS
Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)	31 Avg FPS, 24 1% Low FPS
Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)	40 Avg FPS, 28 1% Low FPS
Framework 16 | RTX 5070	18 Avg FPS, 8 1% Low FPS

Avg FPS

1% Low FPS

MSI Stealth 16 AI+ | RTX 5080 (125 W)

67

39

Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)

80

46

Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)

96

42

Framework 16 | RTX 5070

67

37

0255075100

Baldur's Gate 3 (Native res. Ultra) Data

Product	Value
MSI Stealth 16 AI+ | RTX 5080 (125 W)	67 Avg FPS, 39 1% Low FPS
Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)	80 Avg FPS, 46 1% Low FPS
Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)	96 Avg FPS, 42 1% Low FPS
Framework 16 | RTX 5070	67 Avg FPS, 37 1% Low FPS

Avg FPS

1% Low FPS

MSI Stealth 16 AI+ | RTX 5080 (125 W)

57

49

Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)

75

66

Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)

111

80

Framework 16 | RTX 5070

72

62

037.575112.5150

Black Myth: Wukong benchmarking tool (Native res. High) Data

Product	Value
MSI Stealth 16 AI+ | RTX 5080 (125 W)	57 Avg FPS, 49 1% Low FPS
Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)	75 Avg FPS, 66 1% Low FPS
Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)	111 Avg FPS, 80 1% Low FPS
Framework 16 | RTX 5070	72 Avg FPS, 62 1% Low FPS

Avg FPS

1% Low FPS

MSI Stealth 16 AI+ | RTX 5080 (125 W)

62

43

Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)

63

53

Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)

77

55

Framework 16 | RTX 5070

41

32

020406080

F1 24 (Native res. Ultra High) Data

Product	Value
MSI Stealth 16 AI+ | RTX 5080 (125 W)	62 Avg FPS, 43 1% Low FPS
Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)	63 Avg FPS, 53 1% Low FPS
Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)	77 Avg FPS, 55 1% Low FPS
Framework 16 | RTX 5070	41 Avg FPS, 32 1% Low FPS

Avg FPS

1% Low FPS

MSI Stealth 16 AI+ | RTX 5080 (125 W)

74

48

Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)

68

46

Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)

96

65

Framework 16 | RTX 5070

54

39

0255075100

Metro Exodus Enhanced Edition (Native res. Ultra) Data

Product	Value
MSI Stealth 16 AI+ | RTX 5080 (125 W)	74 Avg FPS, 48 1% Low FPS
Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)	68 Avg FPS, 46 1% Low FPS
Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)	96 Avg FPS, 65 1% Low FPS
Framework 16 | RTX 5070	54 Avg FPS, 39 1% Low FPS

As is the case at 1080p, at 1600p the MSI Stealth 16 AI+ sits between an RTX 5070 laptop and a full-fat (175 W) RTX 5080 laptop in its gaming performance.

At both 1080p and 1600p (native) resolution, the power-limited RTX 5080 in the Stealth 16 AI+ tends to churn out more frames than an RTX 5070 mobile but fewer frames than an unrestricted RTX 5080 mobile. There are exceptions in my testing, though these are most likely due to the CPUs in the different laptops I compared the Stealth to. If you play mainly CPU-intensive games, you might do better with a HX chip (and a thicker laptop) rather than an H one like this.

Real-world gaming performance

This chart shows average and 1% low frame rate results in various tested games at 1600p resolution with Quality upscaling enabled, and frame gen where applicable

Cyberpunk 2077 (Native res. RT Ultra + upscaling (quality), 2x FG)Black Myth: Wukong benchmarking tool (Native res. High + upscaling (quality, 67%))F1 24 (Native res. Ultra High + upscaling (quality), 2x FG)

Avg FPS

1% Low FPS

MSI Stealth 16 AI+ | RTX 5080 (125 W)

97

53

Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)

95

79

Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)

124

83

Framework 16 | RTX 5070

75

58

037.575112.5150

Cyberpunk 2077 (Native res. RT Ultra + upscaling (quality), 2x FG) Data

Product	Value
MSI Stealth 16 AI+ | RTX 5080 (125 W)	97 Avg FPS, 53 1% Low FPS
Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)	95 Avg FPS, 79 1% Low FPS
Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)	124 Avg FPS, 83 1% Low FPS
Framework 16 | RTX 5070	75 Avg FPS, 58 1% Low FPS

Avg FPS

1% Low FPS

MSI Stealth 16 AI+ | RTX 5080 (125 W)

87

74

Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)

84

72

Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)

111

80

Framework 16 | RTX 5070

72

62

037.575112.5150

Black Myth: Wukong benchmarking tool (Native res. High + upscaling (quality, 67%)) Data

Product	Value
MSI Stealth 16 AI+ | RTX 5080 (125 W)	87 Avg FPS, 74 1% Low FPS
Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)	84 Avg FPS, 72 1% Low FPS
Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)	111 Avg FPS, 80 1% Low FPS
Framework 16 | RTX 5070	72 Avg FPS, 62 1% Low FPS

Avg FPS

1% Low FPS

MSI Stealth 16 AI+ | RTX 5080 (125 W)

147

56

Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)

144

125

Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)

184

91

Framework 16 | RTX 5070

92

70

050100150200

F1 24 (Native res. Ultra High + upscaling (quality), 2x FG) Data

Product	Value
MSI Stealth 16 AI+ | RTX 5080 (125 W)	147 Avg FPS, 56 1% Low FPS
Asus ROG Zephyrus G16 (2026) | RTX 5080 (135 W)	144 Avg FPS, 125 1% Low FPS
Lenovo Legion 7i Pro Gen10 | RTX 5080 (175 W)	184 Avg FPS, 91 1% Low FPS
Framework 16 | RTX 5070	92 Avg FPS, 70 1% Low FPS

Again, the MSI Stealth 16 AI+ sits somewhere between an RTX 5070 and full-fat RTX 5080 gaming laptop, although with upscaling and frame gen enabled, it seems to sit a little closer to the latter.

In terms of real-world gaming performance, ie, performance with upscaling and/or frame gen enabled, this laptop performs very well, running not too far behind a laptop with the full-fat RTX 5080 mobile. In terms of actual frame rates, in most games expect well above 60 fps on ultra settings, with ray tracing enabled, at native resolution.

And of course expect even better than this when you lower your game settings down below max, as this will net you frame rates high enough to make good use of that 240 Hz refresh rate. Not to mention lower latency, too, because frame gen does better in terms of latency when the starting frame rate is higher.

Extreme vs Balanced mode

Comparing gaming performance, frequency, and temps for Extreme Performance (+ Apex) mode vs Balanced mode.

Versus Comparison Data

Attribute	MSI Stealth 16 AI+	MSI Stealth 16 AI+ (Balanced mode)
3DMark Time Spy Extreme	8646	7238
Metro Exodus Enhanced Edition (1080p Ultra)	102	89
Gaming temperatures	92	83
GPU stats	2079	1607

MSI Stealth 16 AI+MSI Stealth 16 AI+ (Balanced mode)

VS

MSI Stealth 16 AI+MSI Stealth 16 AI+ (Balanced mode)

3DMark Time Spy Extreme (Overall index score)

+19%8,646

7,238

Metro Exodus Enhanced Edition (1080p Ultra) (Avg FPS)

+15%102

89

Gaming temperatures (Avg CPU (°C))

+11%92

83

GPU stats (Avg GPU freq. (MHz))

+29%2,079

1,607

Putting the laptop into Balanced mode cuts down the GPU frequency and CPU temps, but does cost 15–20% performance in-game.

While the above tests were done with the laptop in Extreme Performance mode, I also tried out the laptop in Balanced mode, which limits frequencies and temps a little and keeps the fans quieter. If you're playing graphically intensive games, I really wouldn't recommend this as the drop in performance is quite significant. It drops frame rates by about 15–20%, turning it more decidedly into something like an RTX 5070 machine, and the fan noise reduction isn't really worth that. But for less demanding games where you don't need that extra 15+%, it's a good shout.

I also tried disabling the 'Apex Mode' which pushes things to the max, even above the 'Extreme Performance' defaults, but I found there to be little in it, just a 1-3% decrease in performance and basically no change in max temps, just 1 °C lower for the GPU.

Overall, I'm pretty impressed with how this Stealth performs given its sleek form factor and pleasant-sounding fans (they do get loud, just not in an obnoxious way). You're getting better performance than an RTX 5070 laptop in a form factor that's usually reserved for more mid-range and less thermally expulsive internals. If you want the best performance possible from a high-end machine, this isn't it, but if you're willing to sacrifice a little for the sleek design, it's stellar.

System performance

The laptops are here tested against intensive benchmarks that mostly test the CPU, but also the GPU, for productivity tasks.

7zip 24.07Cinebench 2024Blender 4.2.0 (junkshop)Procyon

Compressing (GIPS)

Decompressing (GIPS)

MSI Stealth 16 AI+ | Intel Core Ultra 9 386H | RTX 5080 (125 W)

105.399

98.828

Asus ROG Zephyrus G16 (2026) | Intel Core Ultra 9 386H | RTX 5080 (135 W)

117.88

100.88

Lenovo Legion 7i Pro Gen10 | Intel Core Ultra 9 275HX | RTX 5080 (175 W)

162.454

184.259

Framework 16 | AMD Ryzen AI 9 HX 370 | RTX 5070

108.822

128.355

050100150200

7zip 24.07 Data

Product	Value
MSI Stealth 16 AI+ | Intel Core Ultra 9 386H | RTX 5080 (125 W)	105.399 Compressing (GIPS), 98.828 Decompressing (GIPS)
Asus ROG Zephyrus G16 (2026) | Intel Core Ultra 9 386H | RTX 5080 (135 W)	117.88 Compressing (GIPS), 100.88 Decompressing (GIPS)
Lenovo Legion 7i Pro Gen10 | Intel Core Ultra 9 275HX | RTX 5080 (175 W)	162.454 Compressing (GIPS), 184.259 Decompressing (GIPS)
Framework 16 | AMD Ryzen AI 9 HX 370 | RTX 5070	108.822 Compressing (GIPS), 128.355 Decompressing (GIPS)

Multi-core index score

Single-core index score

MSI Stealth 16 AI+ | Intel Core Ultra 9 386H | RTX 5080 (125 W)

1221

124

Asus ROG Zephyrus G16 (2026) | Intel Core Ultra 9 386H | RTX 5080 (135 W)

1215

124

Lenovo Legion 7i Pro Gen10 | Intel Core Ultra 9 275HX | RTX 5080 (175 W)

2039

138

Framework 16 | AMD Ryzen AI 9 HX 370 | RTX 5070

1158

105

07501,5002,2503,000

Cinebench 2024 Data

Product	Value
MSI Stealth 16 AI+ | Intel Core Ultra 9 386H | RTX 5080 (125 W)	1221 Multi-core index score, 124 Single-core index score
Asus ROG Zephyrus G16 (2026) | Intel Core Ultra 9 386H | RTX 5080 (135 W)	1215 Multi-core index score, 124 Single-core index score
Lenovo Legion 7i Pro Gen10 | Intel Core Ultra 9 275HX | RTX 5080 (175 W)	2039 Multi-core index score, 138 Single-core index score
Framework 16 | AMD Ryzen AI 9 HX 370 | RTX 5070	1158 Multi-core index score, 105 Single-core index score

CPU (samples/min)

GPU (samples/min)

MSI Stealth 16 AI+ | Intel Core Ultra 9 386H | RTX 5080 (125 W)

79.43

1691.37

Asus ROG Zephyrus G16 (2026) | Intel Core Ultra 9 386H | RTX 5080 (135 W)

85.5

1642.33

Lenovo Legion 7i Pro Gen10 | Intel Core Ultra 9 275HX | RTX 5080 (175 W)

156.29

1883.93

Framework 16 | AMD Ryzen AI 9 HX 370 | RTX 5070

93.15

1026.08

05001,0001,5002,000

Blender 4.2.0 (junkshop) Data

Product	Value
MSI Stealth 16 AI+ | Intel Core Ultra 9 386H | RTX 5080 (125 W)	79.43 CPU (samples/min), 1691.37 GPU (samples/min)
Asus ROG Zephyrus G16 (2026) | Intel Core Ultra 9 386H | RTX 5080 (135 W)	85.5 CPU (samples/min), 1642.33 GPU (samples/min)
Lenovo Legion 7i Pro Gen10 | Intel Core Ultra 9 275HX | RTX 5080 (175 W)	156.29 CPU (samples/min), 1883.93 GPU (samples/min)
Framework 16 | AMD Ryzen AI 9 HX 370 | RTX 5070	93.15 CPU (samples/min), 1026.08 GPU (samples/min)

Overall index score

Duration (Sec)

Sec/image

MSI Stealth 16 AI+ | Intel Core Ultra 9 386H | RTX 5080 (125 W)

2472

40.44

2.53

Asus ROG Zephyrus G16 (2026) | Intel Core Ultra 9 386H | RTX 5080 (135 W)

2464

40.57

2.53

Lenovo Legion 7i Pro Gen10 | Intel Core Ultra 9 275HX | RTX 5080 (175 W)

2853

35.05

2.19

Framework 16 | AMD Ryzen AI 9 HX 370 | RTX 5070

1789

55.9

3.49

07501,5002,2503,000

Procyon Data

Product	Value
MSI Stealth 16 AI+ | Intel Core Ultra 9 386H | RTX 5080 (125 W)	2472 Overall index score, 40.44 Duration (Sec), 2.53 Sec/image
Asus ROG Zephyrus G16 (2026) | Intel Core Ultra 9 386H | RTX 5080 (135 W)	2464 Overall index score, 40.57 Duration (Sec), 2.53 Sec/image
Lenovo Legion 7i Pro Gen10 | Intel Core Ultra 9 275HX | RTX 5080 (175 W)	2853 Overall index score, 35.05 Duration (Sec), 2.19 Sec/image
Framework 16 | AMD Ryzen AI 9 HX 370 | RTX 5070	1789 Overall index score, 55.9 Duration (Sec), 3.49 Sec/image

The MSI Stealth 16 AI+ keeps up with similar laptops for productivity, though struggles to match the chops of a 175 W RTX 5080 and more powerful Core Ultra 9 275HX.

Productivity-wise, the story is very much the same. It's a high-end laptop, but the CPU that MSI has opted for compromises in raw horsepower to keep things light and efficient enough for the slim design.

The Core Ultra 9 257HX in the Legion 7i Pro Gen10 runs rings around the Core Ultra 9 386H in this laptop in some benchmarks (Cinebench an 7Zip) and performs just moderately better in the rest. The Stealth's Intel chip does outperform the AMD Ryzen AI 9 HX 370 in the Framework 16, though. It's far from a bad chip, just know that you can get better if you're willing to opt for a chunkier and potentially louder laptop.

MSI Stealth 16 AI+ B3W Thermals

Real-world gaming temperatures

Gaming temperature results recorded during multiple Metro Exodus EE runs

Gaming temperatures

Avg CPU (°C)

Max CPU (°C)

Avg GPU (°C)

Max GPU (°C)

MSI Stealth 16 AI+ | Intel Core Ultra 9 386H | RTX 5080 (125 W)

92

100

81

87

Asus ROG Zephyrus G16 (2026) | Intel Core Ultra 9 386H | RTX 5080 (135 W)

72

98

69

77

Lenovo Legion 7i Pro Gen10 | Intel Core Ultra 9 275HX | RTX 5080 (175 W)

94

105

74

79

Framework 16 | AMD Ryzen AI 9 HX 370 | RTX 5070

65

73

68

71

037.575112.5150

Gaming temperatures Data

Product	Value
MSI Stealth 16 AI+ | Intel Core Ultra 9 386H | RTX 5080 (125 W)	92 Avg CPU (°C), 100 Max CPU (°C), 81 Avg GPU (°C), 87 Max GPU (°C)
Asus ROG Zephyrus G16 (2026) | Intel Core Ultra 9 386H | RTX 5080 (135 W)	72 Avg CPU (°C), 98 Max CPU (°C), 69 Avg GPU (°C), 77 Max GPU (°C)
Lenovo Legion 7i Pro Gen10 | Intel Core Ultra 9 275HX | RTX 5080 (175 W)	94 Avg CPU (°C), 105 Max CPU (°C), 74 Avg GPU (°C), 79 Max GPU (°C)
Framework 16 | AMD Ryzen AI 9 HX 370 | RTX 5070	65 Avg CPU (°C), 73 Max CPU (°C), 68 Avg GPU (°C), 71 Max GPU (°C)

The CPU in the MSI Stealth 16 AI+ runs pretty hot, but that's par for the course for an Intel chip. The GPU, however, runs notably hotter than other laptops we've tested. It stays well within safe limits, though.

Both the Intel processor and the Nvidia GPU in the Stealth 16 AI+ B3W can get pretty toasty, whether gaming or doing productivity tasks. That being said, I've definitely seen some Intel chips get hotter. The GPU, on the other hand, reaches higher max temps than others I've seen, but still stays under 90 °C and well within safe limits.

Importantly, the underside of the laptop doesn't get too hot other than right at the back near the screen, behind the stand-off ridge. If you position the laptop right, you can keep that part from touching your legs when you have it on your lap, but if you don't get it right, be prepared to feel the burn. Ditto touching above the Function keys, which gets very hot while the rest of the keyboard area stays just warm.

And when it comes to the fans, these are fantastic (ba dum tsh). They get loud when they're working hard, but no louder than other gaming PCs I've heard when pushing things to the max. The difference here is they have nary a semblance of whine to them, just a pleasant whoosh, like a desk fan on a warm summer's day.

Turning the performance profile down to Balanced, the same timbre is maintained but things obviously get a little quieter. It's not night-and-day, though, and as I said previously, I'd rather get the extra frames. Only if I'm trying to hear audio out of the speakers would I lower things down to Balanced—with headphones on that have a good seal around them, the fans on Extreme Performance mode aren't audible while gaming anyway.

MSI Stealth 16 AI+ B3W Aesthetics

I was immediately surprised by just how sleek and subtle the Stealth 16 AI+ B3W is. MSI's choice here makes for exactly my cup of tea because I like my laptops to be ones I'm not afraid to take out at a cafe or around less gamer-accepting folk (you know the ones).

The 'Stealth' in its name, then, for me means it can fly somewhat under the radar as a gaming machine when I'm out and about (assuming I have the RGB turned off), but reveal its true gaming power when I get home. And in this sense it definitely succeeds, even given the power limitation caveat mentioned in the section above.

It might seem a little greedy of me, given how well it delivers on the understated, professional look, but I think it could have potentially leaned even further in this direction. For instance, it could have got rid of the subtle embossed pattern on the lid entirely, like how Asus does with its TUF A14, another laptop I love.

That being said, the MSI is almost all the way there, and I really can't complain. Subtle metal styling and a giant trackpad just does it for me.

MSI Stealth 16 AI+ B3W Software

MSI's software is nothing to write home about, but it's fine. Just fine, nothing more, nothing less. I like that you get to choose which modules to install, so you can only keep what you need running. For me, that was just the RGB lighting control and performance mode selections. The interface is clean, if not very aesthetically pleasing. It's not the most intuitive to navigate given the different add-on module sections/names, but once you get used to it it's (you guessed it) fine.

Other than that, the laptop came with Norton installed, which is worth uninstalling, in my opinion. I don't see much need for anything other than Microsoft Defender these days, and if I were going to have another antivirus on there I'd go for ESET, but that's a personal choice. Norton let me uninstall it pretty easily, at any rate.

MSI Stealth 16 AI+ B3W Battery life

Battery life

Testing how long the battery lasts in Balanced mode, medium brightness, Wi-Fi disconnected, in the PCMark 10 gaming battery life test.

PCMark 10

Asus ROG Zephyrus G16 (2026)

117

Framework 16

131

Lenovo Legion 7i Pro Gen10

73

MSI Stealth 16 AI+

104

037.575112.5150

Battery life (minutes)

PCMark 10 Data

Product	Value
Asus ROG Zephyrus G16 (2026)	117
Framework 16	131
Lenovo Legion 7i Pro Gen10	73
MSI Stealth 16 AI+	104

The MSI Stealth 16 AI+ lasts a fair amount of time thanks to its efficient processor while in balanced mode and its RTX 5080's lower TDP. It doesn't quite keep up with some other laptops like the Framework 16, however.

Battery life is really helped out by the relatively efficient Panther Lake CPU here. I didn't run out of battery during a workday when in Balanced mode, and in the PCMark 10 battery life test, it fared a lot better than the Legion 7i Pro Gen10 when set to Balanced mode, showing the benefit of the power-limited GPU and more efficiency-focused CPU.

Saying that, it didn't last as long as the HX 370-touting Framework 16, though that is probably due to the GPU rather than the CPU. In all, though, almost two hours of gameplay is a good result, and shows another benefit of the thermal and efficiency focus here.

MSI Stealth 16 AI+ B3W Value

It's hard to know what exactly to compare this laptop with. On the one hand, it could be compared to chunky gaming powerhouses like the Lenovo Legion 7i Pro Gen10. On the other hand, it could be compared to something like the similarly sleek and luxurious Razer Blade.

The best laptop for comparison right now, however, is probably the Asus ROG Zephyrus G16 (2026). Our Andy reviewed that laptop recently, and it shares the same CPU and GPU, although there's slightly less of a power nerf to the RTX 5080 in that laptop (it runs at up to 135 W unless fully manually unlocked to 160 W, compared to 125 W with the MSI).

The configuration with 32 GB of RAM (not tested by us) can be found for $4,400 at Best Buy right now, which is $400 more expensive than the Stealth. But given the high price, that's just an extra 10%, making the two in roughly the same ballpark. And I have to agree with Andy's sentiment here that it's not straightforward to square such a whopper price tag against the likes of cheaper but more performant high-end laptops like the Legion Pro 7i. At the time of writing, you can pick up an RTX 5080 version of that laptop for just over $3,500.

It would be a close call but I might spend the extra $500 on the MSI for its slim chassis and premium, satisfying keyboard and trackpad, plus its fancy OLED. There is, however, a big 'but'. And that is the audio issue I mentioned earlier.

I pretty much only ever use a USB audio device, but if I wanted to play shooters on the laptop using the default audio device, I'd be hesitant to spend so much on this laptop without a fix for the audio latency from MSI. As such, I can't recommend it as clearly as I otherwise would. Certainly not for $4,000.

However, if you're like me and only ever use USB headphones/earphones when you load up shooters and other games that benefit from snappy audio, and if you're happy sacrificing a little performance for portability and a 'fine out in public' aesthetic, it's worth it. Well, for 2026 prices, at least.

It's kinda weird that it took somebody this long to make a Star Wars squad tactics game, but studio Bit Reactor has, for all intents and purposes, knocked it outta da park. Star Wars Zero Company is a superbly executed⁠—if familiar⁠—tactics game, but it's also a new all-timer Star Wars story: I'm talking Rogue One, Knights of the Old Republic, TIE Fighter-tier.

Melt away everything else, and it's got a Terminator skeleton of pure XCOM, the trendsetting tactics series many of Bit Reactor's senior developers cut their teeth on. But the studio struck out from there into more uncharted territory: lavish production values, zoomed-in action game-style exploration, and some truly knockout set pieces and story beats.

Crucially, none of it came at the cost of challenge or crunchy complexity. It feels like Bit Reactor has pitched a big tent where tactics sickos, RPG storytelling fans, and Joe Sixpack cinematic console game-enjoyers can truly find something to love, not just tolerate.

Zero to hero

Zero Company takes place in the twilight of the High Republic, and sees you inhabiting a Commander Shepard-y point of view character, Hawks. Hawks is a disgraced former Republic officer and, at least in my save file, a Zabrak (Darth Maul guy) who delivers every line in a charmingly incongruous cowboy drawl. While Hawks has a largely set character, their gender, appearance, and dress are highly customizable, including your choice of combat class and Bit Reactor's admirable embrace of Star Wars' cantina-going humanoid aliens. I felt a sense of ownership of my Hawks the same way I do with Geralt of Rivia.

The entire cast shares a similar easy charisma, both in your home base crew and recruitable story-centric operators⁠—it's a winning, "family of friends" ensemble straight out of the BioWare playbook. There are a lot of comfortable Star Wars archetypes among them⁠—ponderous-but-loveable alien bruiser, uptight princess, and laconic bounty hunter, to name a few⁠—but, like the comfort zone combat mechanics, they're executed extremely well and belie surprisingly affecting twists.

There is some serious "Aerith dies in Final Fantasy 7" stuff in Zero Company's back half that knocked my socks off, and Bit Reactor went to such great lengths establishing a narrative status quo, the moments that upended it felt like I was getting smacked in the chest.

Even when the story's cruising at a relatively calm clip, your lovable crew can be at each others' throats. The cast has the sense of friction and conflict I'm always looking for in an ensemble, and the ex-Separatist characters are some of my favorites. Former bean counter and Zero Company operations lead Runa is abrasive, smarmy, and unpleasant in the best way, while Umbaran bubble boy Luco hates your guts, as well as those of your clone buddy, Trick.

Zero Company is basically Star Wars A-Team: You're mercs with hearts of gold who do private war stuff for the highest bidder, but in, like, a cool, fun, and ethical way. Galactic Republic Intelligence drops a fat contract in your lap to investigate a kooky cult (it's always a cult), leaving Hawks and crew to turn-based battle their way into saving the galaxy.

You gear up and manage your operation from the Den, a small but well-realized home base in an asteroid-bound smuggler city. From the galaxy map, you can choose to execute from a selection of text-based, noncombat espionage operations reminiscent of the War Table stuff in Dragon Age: Inquisition, while each "Cycle" (read: day) at the Den is capped off with a full-fat combat mission that moves time⁠—and, when appropriate, the story⁠—forward.

You explore the Den with full on, action game-style driving third-person controls, and this is one of Zero Company's neatest tricks. Every environment is as tactile and richly detailed as something in Jedi: Fallen Order, and major story missions even feature exploratory interstitials between the fights. But, as soon as combat starts, Zero Company seamlessly zooms out to a classic, tactical view.

The exploration gameplay is almost painfully simple⁠—interact with points of interest for small bonuses, otherwise you're mostly walking down corridors⁠—but I found it had a really outsized effect in this tactics game context. Even as a first attempt, it adds this revelatory immersive quality, while Bit Reactor managed a level of presentation and wow-factor that I'd love to see in other tactics games and CRPGs.

Squad wipe

You're limited to four squadmates on a mission, and can embark on a variety of odd jobs between main quest blowouts: Hostage rescue, steal the thing, blow the thing up, and the surprisingly rare "kill all the guys." Even on side missions, there's an impish, trolling quality to Zero Company's design that I always love to see.

Enemies will spawn just far enough away that you'll struggle to reach them on a first turn, but they'll have no problem counterattacking after you've committed. Reinforcements often spawn right in your path to an exfiltration, and it seems only enemy snipers have access to the really good rifles that can pinpoint strike you from across the map.

Zero Company's difficulty feels very well-judged to me: I felt pressured enough on Normal, but not overwhelmed. My guys would get downed here and there, but I only had to take a second try on one fight in my playthrough.

I'm a huge sucker for gimmick battles in a turn-based game⁠—the Iron Throne in Baldur's Gate 3, for example, or the flaming oil tower from Divinity: Original Sin 2⁠—and Zero Company's main campaign missions are consistently full of "oh shit" challenges that had me grinning. One of my favorites had me and my squad trooping out in explore mode down a suspiciously deserted catwalk with the MacGuffin of the day at the far end.

Once we grabbed the thing and the trap was sprung, enemies started filling in the path to the exit, and droid bombers began strafing the bridge in a turn-by-turn expanding killzone that pressured me to stay moving even as bad guy reinforcements blocked my path.

In other missions, things could get really weird: One memorable main story level gifted me a one-time, functionally invincible super-unit to clear a path as I had my normal guys follow along and complete objectives⁠—it felt like a reverse-escort mission in this delightfully surprising way. Another jaw-dropper, season finale of a mission hit me with what I can only describe as "Spec Ops: The Line type-shit," a simple-but-devastating mechanical twist that perfectly complimented the narrative one.

Buildy boys

I was surprised at how much RPG buildcrafting there is in Zero Company. We're not quite at Rogue Trader levels of spreadsheet psychosis, but there are some fun synergies to exploit. A little ways in, Zero Company introduces multiclassing, and then we're really off to the races.

My jam has been the Gunslinger: They get stronger the more attacks they make in a round, and also get a quickdraw ability that refunds its action point cost on a kill. Crowds of one-shot, dopey battle droids came to resemble juicy Thanksgiving turkeys to my quickdraw-addled mind, powering me up for DPS attacks on beefier foes.

Astromech droids are support gods in Zero Company, and I almost never went out without one. They can't make direct attacks, but they can refresh squadmates' AP and get bonus slots and uses for utility items like grenades. They're support-casters that feel like College of Lore Bards in Baldur's Gate or the Officer class in Rogue Trader.

In addition to Zero Company's main cast, you can hire generic mercenaries more in line with the anonymous, proc-gen squads of XCOM. I mostly stuck with the canon crew, but I was pleasantly surprised at how seamlessly the custom operators slot into the Den. They draw from a selection of "personalities," each with their own voice actor, and they can be found chilling around the Den like Mother Base soldiers in Metal Gear Solid 5. They even have relevant commentary on the story as it unfolds, and will converse with each other as well as the story squaddies.

Hanging over all of them is the threat of permadeath. I didn't experience its consequences directly in my playthrough, but even the threat of it lends each fight higher stakes⁠. I can't help but admire the chutzpah of hanging permanently missable story content on such a lovable, fragile cast. These systemic risks are also supported by the story in a really clever way: Some characters leave the stage in shocking, permanent fashion. It all contributes to a narrative that feels tragic and grown up in a similar way to Rogue One.

Complete package

Zero Company has been a treat from beginning to end, and I've found little to complain about. There's small stuff, some patchable, and some less so⁠; EA has indicated there will be a day one patch I did not have access to in the review period. My biggest issue: I'm a pretty anti-waypoint guy these days, and Zero Company's exploration segments are so linear, the ever-present Skyrim quest marker felt like it was leading me around by the nose⁠—I'm shocked there isn't already a toggle in the options.

I experienced two hard crashes in my playthrough, as well as one severe graphical glitch that required me to reload a save. Smaller issues popped up here and there: minor graphics hiccups, flickering UI elements, a long hang time while the game would try to resolve an enemy turn, but nothing I found particularly odious. I am sorry to say that Zero Company is not a Steam Deck candidate, and it's hard to see how Bit Reactor could make it fit that box. Bottoming out settings and maxing FSR resulted in a soupy, unparsable, 15-20fps mess for me.

Minor issues aside, Star Wars Zero Company is one of the best games of 2026 and a new all-timer Star Wars videogame. While I don't see it having the modding/sandbox "forever game" gravity of XCOM 2, it is excellent, substantial, and highly replayable. I've found it to be a mic drop of a debut from Bit Reactor, a sense that's only compounded by its crowd-pleasingly anachronistic price tag of $50.

My prime PC gaming hours tend to be when the rest of my household is already in bed. As such, the fantastically clicky-clacky key presses of a mechanical gaming keyboard echoing through our halls isn't really ideal. It's just as well then that Be Quiet! is back with a welcome update to the very nearly silent keyboard I loved last year, the Light Mount.

As someone with a perennially messy desk, the Light Mount TKL's 80% size is just right; I have room to swish my mouse, and I find I'm not forever accidentally hitting the capslock key as with other cut-down keebs. This TKL update offers more than just a smaller form factor, though, introducing the next generation of Be Quiet!'s Silent switches in Linear and Tactile flavours.

This TKL update also addresses one of the key problems I had with the original Light Mount by introducing new gaming features. This includes Quick Tap (SOCD), which will be sweet percussion to the ears of shooter fans—and at best eyebrow raising for the esports sickos. For those unfamiliar, SOCD stands for 'Simultaneous Opposing Cardinal Directions'. Features falling under this umbrella, like Razer's Snap Tap and now Be Quiet!'s Quick Tap, prioritises your most recent directional input, making for strafing so efficient that Valve banned it from Counter-Strike 2.

Alongside Quick Tap, there's also more options for macros, and a few fresh customisation options within the IO center software. While I appreciate these additions and expansions, the TKL still isn't the most feature-packed gaming keyboard you'll find.

I still love the even-quieter switches, though. Coupled with the TKL's sturdy construction, this makes for another smart-looking package. However, it's rocking up to the rowdy PC gaming party as the quiet kid in the corner—that is to say, while some will definitely appreciate its charms, it can't compete with the louder, more popular keeb crowd.

Features

The Light Mount TKL is an 80% keyboard that enjoys the latest generation of Be Quiet!'s Silent switches. Tactile switches are available, but my review unit came with the Silent Linear V2s (though if you're craving clicky-ness, the media knob in the keyboard's top left corner has got you covered). According to Be Quiet!, the new V2s feature thicker gold contact points for the copper lever inside, as well as 'optimised lubrication for quieter operation and improved responsiveness'—well, I'll be the judge of that.

The V1 switches were already astonishingly understated on the full-sized board, making the most of the original Light Mount's internal layers of sound dampening. The V2 switches on the TKL are somehow even quieter.

Though the new switches feel a bit on the firm side right out of the box, they soon actuate just as cleanly as the V1s. Before, I described typing on the Tactile V1s as a controlled descent onto solid ground, but the Linear V2s have a little more pleasing texture underhand. It's not like stepping on a discordantly crunchy leaf, it's more like tiptoeing over a thin layer of soft sand—I like it!

The combination of the V2s and the TKL's construction also means the space bar is now much quieter than what I found on this board's big brother. Although, it's worth noting that future productions of both the full size Light Mount and the Dark Mount will come with the new V2 switches.

So far, so excellent: the Light Mount TKL is an actually quiet mechanical keyboard that won't tattle on you during late night gaming sessions. So, what about those fresh gaming features? Well, let's get the least exciting addition out of the way first: You can now disable the Capslock key in Game Mode.

Cool. Moving swiftly on, the Light Mount TKL also enjoys Be Quiet's version of SOCD, Quick Tap, for the first time. Whereas the original Light Mount felt more like a fairly quiet mechanical keyboard you could also use for gaming, the TKL feels a smidge more like a gaming keyboard in its own right.

And you know what gamers love above all else? Bathing themselves in the glow of RGB lighting. Like its predecessors, the TKL implementation of RGB is similarly classy; a dedicated light strip lines the very top of the keyboard, and every switch below enjoys an 'ARGB diffuser' for tasteful per-key illumination. On top of plenty of customisation options in the IO Center software, the Light Mount TKL also features support for Microsoft Dynamic Lighting.

Performance

The Be Quiet! Light Mount TKL is a joy to type on. Switch-feel is really second to none in my mind. But what about gaming? Well, as I've said before, my most intense encounter in an online game recently was getting lost in Big Walk (not to mention how I tend to spend more time hiding in bushes in Arc Raiders than running and gunning). Similarly, this isn't a gaming keyboard with the competitive scene in mind. With those expectations set, let's get into it.

I tend to accidentally hit the Capslock key on smaller form factor keyboards, such as the Cherry XTRFY K5 Pro TMR Compact. The option to disable this key in Game Mode is definitely welcome, though I do find myself stumbling on to Capslock a lot less with this TKL. I also found this keyboard's layout and 1,000 Hz maximum polling rate does little to impede my running around the feline-themed post office of Cat Mail Co. The Shift key for sprinting between barely organised shelves is comfortably within reach—so long as you don't subconsciously curl your hand into a weird cat paw-esque grip like I do.

The TKL's precision is definitely welcome as I roll up everything in sight in Once Upon a Katamari. The TKL's responsiveness already makes it easy to both slow my roll and steer clear of disaster. Quick Tap comes especially in handy here, making changing the direction of my Katamari all the snappier. Does that mean I start immediately smashing through all of my previous records? Eh, no, not exactly. A swish gaming keyboard can only do so much for a skill issue, after all.

Whether I'm whiling away the wee hours trying to improve my Katamari game, or simply lose track of time playing Paralives, the Light Mount TKL's luscious layers of sound-dampening foam and updated Silent switches continue to keep the secret of my late night gaming sessions.

Build Quality

This is a lovingly put together keyboard. I've already waxed lyrical about the factory-lubricated V2 Silent switches and their lovely, linear actuation. The keys themselves aren't completely smooth on top, offering a little bit of surface texture that's typical of PBT caps. This choice of material is generally more durable than ABS, and also tends to be much more resistant to picking up finger grease. In other words, these PBT keycaps will be looking box fresh for much longer.

Speaking of looks, you're not going to find the Light Mount TKL in any fun colourways. Normally, I would rue the lack of a lavender colourway, but I respect Be Quiet's vision; this is a smart looking keyboard that wouldn't be out of place in either a gamer den or an office with a strict dress code. Besides, if you're still desperate for a splash of colour, the IO Center software offers lots of customisation options for the RGB lighting.

A potentially unfortunate wrinkle is that PBT keycaps aren't known for letting a lot of light through. That means that, even with the keycaps' shine-through legends, the RGB lighting here is generally a more subtle affair. An arched ARGB diffuser on each switch keeps what glow there is evenly distributed, but I can't help but feel the original Light Mount's RGB was a smidge brighter.

Mind you, the gamer den I was testing the fullsize board in is generally much darker than the bright overhead lights of PCG towers where I spent most of my quality time with the TKL. Anyway, the strip of RGB at the top of the TKL is all the lovelier for being allowed to shine unimpeded.

Like the company's other boards, the TKL also enjoys three internal layers of sound-dampening foam. This makes good on the entire promise of Be Quiet! while also making for a keeb with a pleasing amount of sturdy heft. The same brushed aluminium top plate featured on Be Quiet's other boards is also the finishing touch here, acting as the shiny cherry topping an already beautifully constructed cake.

Software

IO Center can either be downloaded as a local desktop app, or accessed via the web. For the most part, it's a smartly laid out bit of software that makes it easy to tweak an array of settings. The latest version offers a helpful hardware monitor too, showing you temps and even fan RPM right there in the dashboard, though obviously this is less relevant in a review about a keyboard.

What is relevant is the robust selection of customisation options for the RGB lights. The usual rippling designs return, allowing for gradient, single, or dual colour lighting effects. If you aren't digging IO Center's 'general' pre-sets, you can create your own layered lighting, customising the look of each key and segment of the top light bar. The Light Mount TKL also enjoys support for Microsoft Dynamic Lighting for those with a much grander, all encompassing vision for their desk.

Now, obviously RGB lighting isn't the only thing that makes a gaming keyboard. Via IO Center, you can also fine tune the Game Mode across a small selection of toggles. These allow you to disable various key configurations while the mode is active, including Shift + Tab, Alt + F4, the Windows Key, Alt + Tab, and now Capslock too.

Interestingly, Quick Tap falls outside of IO Center's Key Binding sub-menu. Instead, you'll find options for remapping keys, editing the Function layer, and recording a whole load of Macros.

Quick Tap is found under 'Product Settings,' alongside the Game Mode. Quick Tap can be calibrated across five key pairs ('A' and 'D' are included as an editable default pair to start you off). While in Quick Tap settings, the software also warns that some competitive titles will restrict or ban SOCD tech, and advises that you check the rules for your game of choice before calibrating.

Value

The Light Mount TKL costs $140/€140. That's about $30/€30 less than the full size version would set you back at launch, though you can get a 100% Light Mount for much less now. That's just as well, as the full size version graces our best gaming keyboards guide as our pick for the best silent board. The obvious question, then, is does the TKL do enough to dethrone its older sibling?

Well, the TKL introduces a number of features that will be coming to future productions of the full size keyboard, such as the V2 switches. Those who already have the full size Light Mount will also be able to enjoy the slightly expanded Game Mode through updates to IO Center, though Be Quiet is still evaluating if further features will be added retroactively.

What I can tell you for sure right now is, due to technical limitations, neither the full size Light Mount or Dark Mount will have the TKL's 10 customisable layers of lighting effects, or its ability to record up to 999 Macro events. The original Light Mount is limited to 58, while the Dark Mount can record up to 200. So, if you use a lot of complex Macros and are looking for a quiet keeb, the TKL is the gaming keyboard to go with.

But that's the thing: many PC gamers will tell you 'the clack is the whole point' of a mechanical keyboard. Personally, I think there's room for a range of sound profiles in this space, but the Light Mount TKL's central reason for being is arguably niche in appeal.

As an 80% design, the TKL is far from the most space-saving thing either, plus plenty of smaller boards beat it on features, too. For instance, our pick for the best 75% keyboard is the Keychron K2 HE, which comes with Hall effect switches and offers customisable actuation points. It's also wireless, woody, and basically the same price as Be Quiet's slightly larger offering.

If you're looking to save yet more desk space, our pick for the best 60% keyboard is the Wooting 60HE v2, which our Jacob describes as "the best keyboard that Wooting has ever made." That's high-praise that feels absolutely earned once you hear just how feature-packed it is.

You can also get a feature-packed 65% gaming keyboard for much less than the Light Mount TKL. Case in point, the best budget gaming keyboard is the Gamakay x NaughShark NS68, and having spent some quality time with this $40 keyboard myself, going with something 100 bucks more feels like a steep ask.

Obviously, a $40 keyboard cannot hold a candle to Be Quiet's build quality, and its software also leaves a lot to be desired compared to IO Center's ease of use. But otherwise the NS68 still offers fantastic value for money with Hall effect switches and even PBT keycaps.

As a typist, I still love the Light Mount TKL; the switches really are that quiet and the key actuation is wonderfully smooth. Though I rue the lack of fun colourways, the aluminium topped chassis definitely looks very classy as it is, and I appreciate all of the options for colourful RGB customisation in the IO Center software. Although not the brightest, the RGB here illuminates my dank gamer den well enough.

As a PC gamer, I am left wanting just a little more for the money—specifically, the option to customise key actuation points wouldn't go amiss. To be fair, having seen how much the full size Light Mount has come down in price since its launch last year, the TKL may also drop to a similarly appealing price point before too long.

Furthermore, I'm hardly a high-tier competitive gamer and what's here on the TKL definitely suits most of my gaming needs. If you can take or leave gaming keyboards' bleeding edge features, but still want a well-made mechanical keyboard to class up your desk, there's lots to like about the Light Mount TKL.

Blood Dungeon adheres rigidly to the autoshooter or 'bullet heaven' format: it's about moving around a scrolling arena auto-gushing projectiles that turn a steadily increasing horde of enemies into XP. Every level up grants a choice between two or three randomised perks or improvements, steadily turning the player into a blur of pulsating DMG-emitting fireworks. Rounds are usually over in less than ten minutes.

I loved Vampire Survivors but then I got over it. I've never been gripped by any of the clones. But Blood Dungeon has some appealing bullet points going for it: it transforms the commonly top-down autoshooter into a sidescrolling platformer. At a glance, it looks more like Spelunky than it does Brotato or Deep Rock Galactic: Survivor.

The presentation is more appealing than your average bullet heaven fare too: Blood Dungeon is the work of Nidhogg studio Messhof and their distinctively globular 2D art style is a perfect fit for precision-oriented action. Animations in particular are often brilliant, and the music—which its creator Matthew Hoey aptly describes as dungeon drum 'n' bass—is simultaneously madcap and hypnotic.

There's no denying Blood Dungeon is a sleek, well-wrought take on this modern subgenre. It's probably among the best examples of it. But as the hours I spent with it wore on it started to feel exquisitely meaningless. I was reminded why I dropped Vampire Survivors.

Blood Packed

Blood Dungeon starts like all roguelites: vainly. I am Gun Man, a man with a piddling machinegun which shoots every 1.21 seconds. Thanks to the blood left by my fallen enemies, within a minute I've upgraded it to shoot twice rather than once, and then, thirty seconds later, I've acquired a sword that rotates around my body, shredding bats as I roam the lurid halls. By the time I die there are four double-sized swords surrounding me, my gun issues five bullets per round virtually non-stop, and the halls have turned into impenetrable walls of monster flesh.

Five hours later I've unlocked a handful of other characters and all six of the maps. I've upgraded my health, my jump height, my damage output, the amount of weapons I can equip per run, and my repertoire of powerful upgrades has expanded dramatically. If I'm lucky my character will be lost among an overwhelming array of erupting, tantruming firepower by the eighth minute.

My favourite weapon for achieving this is easily the bouncing ball, which can demolish most of the game's first bosses within seconds if I focus on upgrading it. I also love the turrets and the laser, though the shotgun whose spread can be increased from three to upwards of 13 shots is pretty fun too.

I also like most of the characters, except the unwieldy flying mosquito (I'll admit it's a skill issue). They all throb with panic, which well they should. My overall favourite is probably Skinja, who cartwheels around the maps at a blinding pace while issuing volleys of the aforementioned shotgun pellets. I also liked the Baby, whose low health is well-compensated by a revolver that one-shots most early round enemies. There's also Hypnomancer, who loads out with gas which, naturally, he farts from his arse.

But it's the sidelong perspective that appeals to me most, even though all maps stick to common survivor archetypes: there's the X-axis corridor map, the Y-axis pit map, and other gimmicks like the map with a lot of water, and the map you have to mine through. Importantly, every map has a fixed layout but the locations of items and keys are, for the most part—but not entirely—randomised.

Most maps also loop back on themselves, so the genre's sense of constant, hysterical forward momentum is complemented by a newfound focus on map knowledge: knowing where to reliably find health and explosives is the difference between beating the area's third and final boss or dying for the umpteenth time to its second. If I spend the less frantic opening minutes unlocking as many doors as possible, it'll make the final minutes when there's barely a safe place to stand much easier.

Slot Your Throat

But Blood Dungeon does inevitably feel like a grind. The opening hours are an enticing onslaught of rewards and unlocks, but by the ten hour mark, with my fist deep in its basket of candy, the basket is ripped away and my nose pressed to the grindstone. It lays bare the disappointing truth that the game itself can't withstand the absence of its rewards.

Still, Blood Dungeon feels superb to play which is essential for an arcade game but oddly, kinda rare for a Survivors-like. I guess most Survivors-likes don't need to feel good because the genre usually amounts to little more than a slot machine simulator. It fast-tracks RPG character progress so that dopamine hits come thick and fast in units of seconds rather than hours. Then, in the final five seconds or less, assuming the RNG hasn't been so kind as to make me unstoppable, it steals everything away and invites me to start again.

I can get better and better at moving around in Blood Dungeon but if the RNG doesn't favour me I'm still stuffed. Being at the mercy of chance is baked into its design, of course, but there is a big unambiguous line between the story-generating unpredictability of a Nethack or Spelunky and the more-ish, non-nutritive unpredictability of a lottery.

And that's ultimately what Blood Dungeon and its ilk feel like to me, once I've put the controller down and walked away from their exquisitely compulsive loops and dazzling visual noise. It's addictive, and that's the only reason I keep playing it. It gets easier as I play, but that's not because I've improved much: it's overwhelmingly attributable to the wider array of weapons, charms and spells I have at my disposal at hour fifty versus hour one.

Still, Messhof pulls its own Survivors-like off with flair and applies a thin layer of skill reward. I enjoyed the time I spent playing it. But now that I've put it down, I never want to think about it again.

I've slowly come to the realisation that, for computer speakers, once you get past the mid-range price point it becomes all about diminishing returns. There are some really solid budget picks (with Creative's Pebble Pro being our budget speaker of choice), but the $100-200 range is where you get the most features and the best price-to-value ratio. Go beyond that and you're spending a lot of money for only incrementally better sound and some non-essential luxuries.

And nothing quite proves that better than the Creative XF1, a set of compact yet incredibly useful speakers with a whole host of neat extra features. And they even come with two cork stands to prop the speakers, which can go entirely unused if you like your speakers flat, but It's still plain neat that they come in the box for no extra charge.

Naturally, the more compact you go, the less room you have to jam speakers full of large tweeters and woofers, and as such, the balance with the XF1 isn't just where to cut costs in each unit, but just how small you can condense them down before the sound starts to take a hit.

And Creative, for the most part, has nailed that balance, with a set of speakers that are small enough to fit on my desk, but loud enough to shake it.

Features

Creative XF1 is loaded with great little design decisions. For one, it has a robust set of connectivity options, with it working via USB Type-C, aux, and Bluetooth. And that Bluetooth 6.0 connectivity even supports LDAC wireless, so it retains a high level of quality no matter the connection.

The single button on the front of the right speaker can not only be spun to adjust volume, but also clicked to swap between the connectivity modes. This means swapping from my gaming PC to my phone or laptop requires no more than a mere button press. And, in practice, this works really well too. With my default audio setting being the speaker, I can click to connect to Bluetooth, and my rig swaps to my plugged-in headphones, which is effectively a poor man's version of the NZXT SwitchMix (and being much cheaper, 'poor man's' is apt).

It's a pretty simple set of speakers otherwise. Simply plug it into the mains, connect it to the left speaker, plug it into a device and start playing sound. One neat feature that comes in the box is a set of cork stands. Without clicking into place or requiring any tools, you merely place the speakers on top and can elevate the speakers at a 15° angle. Generally speaking, being able to elevate the speakers, and therefore elevate the tweeter and woofer, means that it directs the sound more towards you. You may not use the stands, but the fact that they come in the box for free is a solid extra.

Sound

The XF1 speakers have striking quality and clarity for the price and size. Bass, mids, and treble come through clear and are easy to pick out in the mix. And it offers strong sound reproduction at low volumes, which some speakers struggle with (here's looking at you Edifier G2000 Pro). They can also get fierce loud, which is likely linked to the fact that they are 72 watts RMS. Even the Kanto Ora, a significantly more expensive set of desktop speakers, only offers an RMS wattage of 50 watts.

Naturally, you will get better-sounding speakers if you spend more (like the Kanto Ora I just mentioned), and if you are willing to dedicate the space for it, you can get truly superb sound quality out of the likes of the Mackie CR3.5BT. But I have yet to find a set of speakers that offers this level of quality out of its size and price.

It's a speaker designed for games, music, and movies, and it handles them all well thanks in part to smart EQ modes in its software. The gaming mode cuts a tad from the upper mids, whilst giving a light boost to everything else, whereas the music mode drops bass and raises sub bass and upper mids. The Movie EQ also cuts the bass somewhat, while raising everything else.

There's a rather impressive amount of EQ options in Creative's software, including EQ presets for specific games. The GTA preset is higher on the upper mids, where the Last of Us provides a boost in bass and upper mids, with light cuts in the lower mids, sub bass and highs. I love how varied these presets are, and I think the speakers are at their best with a very slight uptick in the bass. Still, even at its default preset, the XF1 is a tiny force to be reckoned with.

The metallic clangs of Abyssus are as metallic and clangy as you'd want them to be here. It pounds when you shoot the shotgun, but it's able to catch the nuance of a soft hum or whir. Abyssus is a game all about its murky, watery atmosphere, and that atmosphere is unhindered coming from the light frame of XF1.

For something a bit more aggressive, I sicced the speakers on speedrunner Echobreaker, and found its combinations of synth and techno drums to come through well. When the thumps, the speakers do a good job of representing that thump, shaking on top of the desk.

Moving over to something more overtly musical, Sea Power's Disco Elysium soundtrack sounds great on the XF1. Trumpets in the middle and high end have real depth to them, where strings and piano fill out the bottom of the mix. There's a solid clarity to everything, making it easy to pick out each instrument in their own registers.

You aren't getting audiophile-level quality here, but I also wouldn't expect it. The bass, mids, and highs are all clear and easy to pick out. You can get speakers which offer a deeper sound, you can get speakers that thump a little more when the bass kicks in, and you can hear more nuance in the mix from something more high-end, but what Creative have done for the price is genuinely quite impressive.

Ergonomics

With the width being just under 10 centimetres and length 19 centimetres, these bookshelf-style speakers are light, easy to set up, and can fit on many a desk. Being front-facing, they are clearly designed to sit on top of a desk and point at you from just a few feet away. The size and setup do make them at their best when plugged into a gaming rig, and the included cork stands help carry that sound straight to your ears.

Though it has many connectivity options, you only really need to plug in the power cable and connecting cable between the speakers to get it up and running. That makes it a solid pick in a clean setup with few wires. The benefit of their size and simplicity is that they're easy to move around and even easier to clean. The only thing you have to worry about when wiping down the XF1 is accidentally moving the volume knob and blasting yourself with The Cardigans, or whatever equally banging band you're listening to.

Aesthetics

The review model I got my hands on is the black variant of the XF1, and its looks are simple and understated. There's a small light near the volume knob to show what device you're connected to, but it has no other lighting, and it has a very simple Creative Logo on the bottom of the left speaker. Otherwise, it lacks any branding and could fit into a tidy setup with ease.

On the whole, I'd say an understated aesthetic serves these speakers well. They're pretty no-nonsense in a lot of ways, but can really deliver in the sound department when you crank them up.

Value

Though it doesn't sound as good as the $350 Kanto Ora I tested back in 2024, it offers a very similar experience in a lot of ways. They're both light and small bookshelf speakers with strong connectivity, and they both have an all-black, understated look, making them great in gaming and non-gaming setups.

For something a little closer to their price point, they blow the Edifier G2000 Pro out of the water in both look and sound. Though the G2000 Pro's throwback is more unique, I think Creative wins in looks, sound clarity, and software.

Though they're more expensive than the Logitech Z407 (at around $120), which is another set of mid-range speakers, I'd be picking the XF1. The Z407 does come with a subwoofer, offering bigger bass, which is a huge boon, but sound quality, connectivity, and ease of use are top-notch here. The XF1 also comes with a subwoofer output jack, so you can plug in your own subwoofer if you have one.

If you just need something, I wouldn't go for the XF1. In fact, there are a few cheaper Creative speakers I'd be picking if my budget were tighter, like the Pebble Pro or the Sound Blaster GS3. Still, if you're looking at around the mid to high $100s, there's nothing else I'd be putting on my small desk.

The Creative XF1 are firmly a mid-range set of speakers, with a mid-range price to match, but you are getting a lot of features and clear sound in return.

Stealth has changed.

It's no longer about avoidance, visibility, or tension. It's an endless series of "stealth action" games trivializing the uncertainty of sneaking and pushing players into open conflict. The age of camouflage has become the age of tall grass: stealth is no longer treated as a genre worthy of big budgets, but an optional feature of conventional action games. [Takes a long drag from a cigarette] Stealth has changed. And when the player's visibility is under total control… espionage becomes routine.

Metal Gear Solid 4: Guns of the Patriots was at the vanguard of the stealth action shift in 2008—but playing it today, the particular way it mixes mercurial stealth and heavy weaponry is anything but routine. Snake is outfitted with a tranq pistol, non-lethal CQC training, and the latest OctoCamo sneaking suit, an ingenious invention of espionage that automatically camouflages the user to their surroundings. In his other pocket are enough guns to arm a strike force—SMGs, shotguns, ARs, RPGs, a rail gun—customizable with Create-a-Class attachments.

OId Snake skulks through active battlefields, bewildered by the growing mundanity of war, as Kojima grapples with the undeniable appeal of getting in the shit. This is a stealth game that doesn't mind if you try to play it like Call of Duty, but going loud comes at the cost of stressing Snake the fuck out.

This Master Collection port, sold alone or alongside Peace Walker, is a genuinely thrilling object: proof that MGS4 wasn't a mass hallucination we all had in 2008, but a uniquely strange meditation on existential anxiety—of warmongering doomerism, automation, desensitized violence, and back pain—trapped for 18 years on a PlayStation 3. Konami has done good work here, though at times the bare minimum: you can now play MGS4 at 4K (yay!), but at a locked 60 fps (less yay).

Still, that's a lot better than the extremely unreliable 30 fps it once struggled to maintain, and the Master Collection launcher is less cumbersome now than when Volume 1 arrived on PC in 2023. Guns of the Patriots has gone so ignored for so long that it's hard to split hairs here—this is the new best way to play one of my favorites.

Naked sun

Man does it feel great to see Guns of the Patriots cleaned up. It's a special one for me: MGS4 landed just as I reached the age to enjoy games on a deeper level. It was also my first Metal Gear, which is a little hilarious in hindsight, given how obsessed the story is with its own past.

In 2026 I'm seriously charmed by the 2008-ness of this world: from the fully functional iPod (still intact here, but lacking Konami's official Metal Gear hype podcasts) to Kojima's infatuation with the XM8 and P90 (objectively the coolest guns ever made). Hindsight hasn't done wonders for its meandering plot and sometimes dull cutscenes, but Kojima's vision for a world consumed by greed and accelerated by decision-making AIs is uncomfortably spot-on. More than anything else though, MGS4's long exile has given me a new appreciation for its extremely balanced stealth.

Acts 1 and 2—where MGS4 really flexes its sneaking muscles—are the sort of triumphant stealth sandboxes that are quite rare these days. Maps are focused and linear, but brimming with options. Guards are reactive, consistent, emotive, and learnable. The years have gifted me patience for stillness, and MGS4 is most fun at a snail's pace. Slithering between guards six inches away from their boots camouflaged as a patch of dirt is a bigger thrill than crouching through tall grass in Horizon, Ghost of Tsushima, or any modern Assassin's Creed.

I love that Kojima can't help but hand players a box of toys with hidden functions. I'm always picking up three times more gear than I'll ever use, which I've learned is a blessing, because every replay is an opportunity to get fixated on a tool. This time, I finally learned how useful Playboy magazines can be when you're trying to ghost an area. This is also the playthrough in which I remembered that the Mark 2 robot, a drivable drone with a cloak and stun prod, is very useful for covertly retrieving items in risky locations.

Tactical espionage action

Replaying Guns of the Patriots and Splinter Cell: Chaos Theory in the same year is highlighting just how bland open worlds have made level design. I love MGS5, but its small Far Cry villages that have to be approachable from all angles are trounced by MGS4's busy battlefields. My favorite is the Act 2 raid on Liquid's mansion, where Snake has to bypass several active warzones either by choosing a side or concealing himself in the chaos.

In my first run through the forest, I crawled up the middle until I found a perfectly Snake-sized crevice in a berm that wrapped around the riskiest sightlines. After that I got caught, shot to death, and tried again, this time finding an alternate route by diving under a shallow river. These areas pack a lot of intentional, almost invisible routes in relatively small spaces, and they're patrolled by nuanced guards that inspire experimentation.

PMC troops can hear you but not see you, or see you but not clearly enough to investigate, or be prompted to investigate, but not be in a hurry about it. Their physical and verbal reactions communicate their alertness without today's explicit "detection meters," and Kojima's lack of restraint means there are like 12 different ways to knock one out. It's hard not to think Snake is the coolest spy in the world once you've gone the extra mile to understand MGS4's extra CQC moves—body slams, chokeholds, stun knife zap, disarms, and of course, holding a guy up at gunpoint, searching him for valuables, then squeezing his nuts until he passes out.

This is true of every Metal Gear, but it's so smart to have an "evasion" phase between "caution" and "alert" where guards frantically search the area expecting to find you. It makes stealth far less binary, and the unpredictability of search patterns in this state is a solid punishment for recklessness.

Movie mode

MGS4's enduring flaw is its second half, where it all-but abandons what it does best. Act 3 is a boring follow mission capped off by the greatest "turret sequence" ever conceived. Act 4 is a walk down memory lane that's more button pressing than sneaking, and Act 5 gets brilliant again, but ends too soon.

This sequel is famous for its long cutscenes, exposition dumps, and inscrutable proper nouns, but honestly, I love all that garbage. I'm the guy who finished MGS4 then returned to the launcher to read the MGS4 Database, a contained wikipedia for all things nanomachine. I have a high tolerance and admiration for Kojima's tastes and always find a lot to like about his characters and dramas, cringe be damned. Like how Meryl hates her dad Roy Campbell because she thought he was her uncle her entire life—forgot that bit until a database deep dive. And that weird hairless monkey that drinks cola and smokes cigs? Betcha didn't know his name is Little Gray, and that he only smokes because Snake made it look cool.

For the first half, MGS4 has a damn-near perfect balance of game and spectacle. Then the game parts get shorter and boring. From then on it only occasionally recovers its pulse with a cool boss fight, "holy shit" cinematic trick that other videogames just never even try, and a memorable finale. There is, weirdly, not enough filler. It's enough to make me wonder if MGS4 was squeezed for time and heavily cut in the same way that Phantom Pain would eventually release without a real ending.

I want more video settings in this PC port, that's for sure, but I appreciate the goodies that this Master Collection could've gotten away with skipping, like a digital art book, manual scan, and that sweet database app.

I also clocked some small touches that confirm Konami didn't just boost the resolution and call it a day: There's a late-game moment where you need to shake the PS3 controller that could've been replaced by a generic controller glyph, but instead it recognized I was on a Steam Deck and showed custom art for the machine. And the Codec call where Otacon jokes about putting in "disc two?" He now has new lines about being on the "PC platform, a digital title!" Nice one, Konami.

This game rocks. It's not a bulletproof recommendation given the slog of a back half, but in exchange you get some of the strongest stealth encounters in videogames and Kojima's signature goofy, genuine storytelling with all the glitz that a PS3 could muster (and, blessedly, none of the slowdown). There's still nothing like it 18 years later, and now I can easily revisit it whenever I want.

I'm trying not to be too down on the new(ish) AMD Ryzen 7 7700X3D. After all, we are living in a time when the regular narrative is that all hardware is brutally overpriced and every tech company out there is trying to make a quick buck off the back of an AI bubble that is ever more certain to wipe out the global economy in one big pop. But here's AMD trying to give us an even more affordable AM5 gaming processor with a lovely dollop of the 3D V-Cache good-stuff. And yet I'm still not happy.

I guess, fundamentally, I don't get it. You see, what we have here is a slightly slower Ryzen 7 7800X3D CPU with slightly lower MSRP. On the surface, that sounds great. The 7800X3D was the best gaming CPU of its generation, and only dropped off our recommended list when its Zen 5 replacement, the Ryzen 7 9800X3D, dropped in price.

The issue is that the 7800X3D is old now. I reviewed it—very favourably I might add—back in April 2023. That in itself is not an issue; the underlying silicon is still excellent and still performs fantastically well up against the latest from either AMD or rivals Intel. The issue is that it's been out in the market a long while, most enthusiasts have their eyes on that 9800X3D (or even its slight upgrade, the Ryzen 7 9850X3D), and the ol' 7800X3D still has plenty of numbers out in retail, too. For essentially the same price as this 'new' 7700X3D. Sometimes cheaper.

And therein lies the friction. Why isn't AMD just making more 7800X3D chips and giving them a permanent price cut? Does it really have so many 3D V-Cache'd Zen 4 CCDs lying around—that couldn't quite get to the right clock speeds—that the only way to ship them would be to have them clocked lower out of the gate under a different chip name? And if so, wouldn't you rather spend the same money on the older, slightly faster, theoretically more reliable, CPU?

I kinda feel like I would.

AMD Ryzen 7 7700X3D features

If you remember the old Ryzen 7 7800X3D, then you know exactly what we're talking about here, because this is essentially the exact same silicon, except this time clocked a little lower out of the box.

That headline difference translates to a base clock of 4 GHz for the 7700X3D and 4.2 GHz for the 7800X3D, and then a boost clock of 4.5 GHz versus 5 GHz. That's where the differences begin and end, even when we're taking pricing into consideration, too. The starting price for the 7700X3D is $329, but while the 7800X3D launched with a $450 MSRP that's not its cost right now. There are so many of the excellent old chippies out in retail that the price for a 7800X3D on Amazon right now is just $334; just $5 more than this new budget option.

Otherwise, we're still talking about a Zen 4 CPU, using a single eight-core, 16-thread CCD with a 64 MB dollop of 3D V-Cache slapped on top of it. That layering, and the insulating nature of having some extra silicon sitting between the CPU cores and the CPU cooler, means the Zen 4 iteration of 3D V-Cache necessitated a drop in clock speed. This is why the 7700X3D has a boost clock of 4.5 GHz while the 7700X, with no extra L3 cache, has a maximum boost clock of 5.4 GHz.

Zen 4 itself was, in large part, a respin of the Zen 3 architecture, but with the CCD built on a TSMC 5 nm production process and the IOD on a 6 nm one. It was also AMD's first native PCIe 5.0 and DDR5 platform, too. It got a little more cache, and with the X3D chips a lot more cache. Though the essential design remained more or less unchanged compared with the Ryzen 7 5800X3D which kicked it all off.

The difference between the Zen X3D chips and the Zen 5 X3D chips, however, is in where the extra cache sits. The likes of the 9800X3D and 9850X3D don't have to compromise on clock speed because the 3D V-Cache layer sits beneath the CCD instead of between it and the cooler.

AMD Ryzen 7 7700X3D performance

You will probably not be surprised to hear the performance you should expect to see from the Ryzen 7 7700X3D isn't a big surprise. This is, after all, an eight-core Zen 4 chip with an extra 64 MB of L3 cache bolted onto it. Just like the Ryzen 7 7800X3D. So yes, gaming performance and productivity metrics are pretty damned similar.

There are some differences, though those are minimal and largely down to either the slight difference in clock speed out of the box or simple testing variance.

When it comes to gaming you're looking at 1080p performance that is around 94-95% that of the top of the class, the Ryzen 7 9800X3D of Zen 5 fame, and that means frame rates at the same sort of levels as the 7800X3D. Which, in short, means you're getting excellent gaming performance from this affordable CPU, even if it's not performance we haven't seen before.

Previous

1 / 2

Next

1080p gaming performance

We test games at 1080p to ensure variance between CPU performance is captured

Cyberpunk 2077 (1080p RT Ultra + DLSS Balanced)Baldur's Gate 3 (1080p Ultra)Homeworld 3 (1080p Epic)Metro Exodus Enhanced Ed. (1080p High)Total War: Warhammer 3 (1080p High)

Avg FPS

1% Low FPS

AMD Ryzen 7 7700X3D

112

73

AMD Ryzen 7 7800X3D

120

88

AMD Ryzen 7 7700X

107

79

Intel Core Ultra 5 250K Plus

118

93

AMD Ryzen 7 9800X3D

112

76

Cyberpunk 2077 (1080p RT Ultra + DLSS Balanced) Data

Product	Value
AMD Ryzen 7 7700X3D	112 Avg FPS, 73 1% Low FPS
AMD Ryzen 7 7800X3D	120 Avg FPS, 88 1% Low FPS
AMD Ryzen 7 7700X	107 Avg FPS, 79 1% Low FPS
Intel Core Ultra 5 250K Plus	118 Avg FPS, 93 1% Low FPS
AMD Ryzen 7 9800X3D	112 Avg FPS, 76 1% Low FPS

Avg FPS

1% Low FPS

AMD Ryzen 7 7700X3D

122

52

AMD Ryzen 7 7800X3D

118

77

AMD Ryzen 7 7700X

88

61

Intel Core Ultra 5 250K Plus

109

67

AMD Ryzen 7 9800X3D

146

76

Baldur's Gate 3 (1080p Ultra) Data

Product	Value
AMD Ryzen 7 7700X3D	122 Avg FPS, 52 1% Low FPS
AMD Ryzen 7 7800X3D	118 Avg FPS, 77 1% Low FPS
AMD Ryzen 7 7700X	88 Avg FPS, 61 1% Low FPS
Intel Core Ultra 5 250K Plus	109 Avg FPS, 67 1% Low FPS
AMD Ryzen 7 9800X3D	146 Avg FPS, 76 1% Low FPS

Avg FPS

1% Low FPS

AMD Ryzen 7 7700X3D

104

56

AMD Ryzen 7 7800X3D

101

56

AMD Ryzen 7 7700X

74

40

Intel Core Ultra 5 250K Plus

98

56

AMD Ryzen 7 9800X3D

117

55

Homeworld 3 (1080p Epic) Data

Product	Value
AMD Ryzen 7 7700X3D	104 Avg FPS, 56 1% Low FPS
AMD Ryzen 7 7800X3D	101 Avg FPS, 56 1% Low FPS
AMD Ryzen 7 7700X	74 Avg FPS, 40 1% Low FPS
Intel Core Ultra 5 250K Plus	98 Avg FPS, 56 1% Low FPS
AMD Ryzen 7 9800X3D	117 Avg FPS, 55 1% Low FPS

Avg FPS

1% Low FPS

AMD Ryzen 7 7700X3D

148

78

AMD Ryzen 7 7800X3D

150

86

AMD Ryzen 7 7700X

147

74

Intel Core Ultra 5 250K Plus

145

79

AMD Ryzen 7 9800X3D

148

91

Metro Exodus Enhanced Ed. (1080p High) Data

Product	Value
AMD Ryzen 7 7700X3D	148 Avg FPS, 78 1% Low FPS
AMD Ryzen 7 7800X3D	150 Avg FPS, 86 1% Low FPS
AMD Ryzen 7 7700X	147 Avg FPS, 74 1% Low FPS
Intel Core Ultra 5 250K Plus	145 Avg FPS, 79 1% Low FPS
AMD Ryzen 7 9800X3D	148 Avg FPS, 91 1% Low FPS

Avg FPS

1% Low FPS

AMD Ryzen 7 7700X3D

152

77

AMD Ryzen 7 7800X3D

159

127

AMD Ryzen 7 7700X

156

124

Intel Core Ultra 5 250K Plus

174

141

AMD Ryzen 7 9800X3D

152

79

Total War: Warhammer 3 (1080p High) Data

Product	Value
AMD Ryzen 7 7700X3D	152 Avg FPS, 77 1% Low FPS
AMD Ryzen 7 7800X3D	159 Avg FPS, 127 1% Low FPS
AMD Ryzen 7 7700X	156 Avg FPS, 124 1% Low FPS
Intel Core Ultra 5 250K Plus	174 Avg FPS, 141 1% Low FPS
AMD Ryzen 7 9800X3D	152 Avg FPS, 79 1% Low FPS

You can see the benefits of the 3D V-Cache chips in games versus the higher-clocked 7700X without it, but also just how close the 7800X3D and the 7700X3D are in frame rate terms.

Synthetic gaming performance

We run the industry standard 3DMark test to get a guage of general gaming CPU performance

3DMark Time Spy

CPU index score

Overall index score

AMD Ryzen 7 7700X3D

12848

16435

AMD Ryzen 7 7800X3D

13078

16521

AMD Ryzen 7 7700X

13869

16670

Intel Core Ultra 5 250K Plus

19987

17567

AMD Ryzen 7 9800X3D

16116

17052

3DMark Time Spy Data

Product	Value
AMD Ryzen 7 7700X3D	12848 CPU index score, 16435 Overall index score
AMD Ryzen 7 7800X3D	13078 CPU index score, 16521 Overall index score
AMD Ryzen 7 7700X	13869 CPU index score, 16670 Overall index score
Intel Core Ultra 5 250K Plus	19987 CPU index score, 17567 Overall index score
AMD Ryzen 7 9800X3D	16116 CPU index score, 17052 Overall index score

The extra multi-threaded chops of the Intel chip are starkly highlighted here, even if they don't necessarily make a huge difference in games.

The benefits of the 3D V-Cache lump bonded to the 7700X3D's CCD, however, are most clearly demonstrated by the differences in gaming performance between this newer chip and the older Ryzen 7 7700X.

Just take a look at the frame rate delta between those chips in the particularly cache-sensitive Homeworld 3 and Baldur's Gate 3 benchmarks. The straight 7700X, despite its considerably higher clock speed, is considerably slower than X3D model. An easy example of where fighting over a couple hundred megahertz isn't a battle that PC gamers really need to get into.

Compute performance

We may be PC Gamer, but we also want to know how our CPUs perform in general compute tasks, too

Cinebench 2024Blender 4.2.07zipHandbrake 1.8.1

Single-core index score

Multi-core index score

AMD Ryzen 7 7700X3D

102

1011

AMD Ryzen 7 7800X3D

113

1082

AMD Ryzen 7 7700X

118

1104

Intel Core Ultra 5 250K Plus

138

1817

AMD Ryzen 7 9800X3D

133

1307

Cinebench 2024 Data

Product	Value
AMD Ryzen 7 7700X3D	102 Single-core index score, 1011 Multi-core index score
AMD Ryzen 7 7800X3D	113 Single-core index score, 1082 Multi-core index score
AMD Ryzen 7 7700X	118 Single-core index score, 1104 Multi-core index score
Intel Core Ultra 5 250K Plus	138 Single-core index score, 1817 Multi-core index score
AMD Ryzen 7 9800X3D	133 Single-core index score, 1307 Multi-core index score

AMD Ryzen 7 7700X3D

79

AMD Ryzen 7 7800X3D

81

AMD Ryzen 7 7700X

85

Intel Core Ultra 5 250K Plus

132

AMD Ryzen 7 9800X3D

101

Blender 4.2.0 Data

Product	Value
AMD Ryzen 7 7700X3D	79
AMD Ryzen 7 7800X3D	81
AMD Ryzen 7 7700X	85
Intel Core Ultra 5 250K Plus	132
AMD Ryzen 7 9800X3D	101

Compressing (GIPS)

Decompressing (GIPS)

AMD Ryzen 7 7700X3D

115

124

AMD Ryzen 7 7800X3D

118

128

AMD Ryzen 7 7700X

110

132

Intel Core Ultra 5 250K Plus

147

146

AMD Ryzen 7 9800X3D

132

141

7zip Data

Product	Value
AMD Ryzen 7 7700X3D	115 Compressing (GIPS), 124 Decompressing (GIPS)
AMD Ryzen 7 7800X3D	118 Compressing (GIPS), 128 Decompressing (GIPS)
AMD Ryzen 7 7700X	110 Compressing (GIPS), 132 Decompressing (GIPS)
Intel Core Ultra 5 250K Plus	147 Compressing (GIPS), 146 Decompressing (GIPS)
AMD Ryzen 7 9800X3D	132 Compressing (GIPS), 141 Decompressing (GIPS)

AMD Ryzen 7 7700X3D

70.8

AMD Ryzen 7 7800X3D

76

AMD Ryzen 7 7700X

79

Intel Core Ultra 5 250K Plus

101

AMD Ryzen 7 9800X3D

92

Handbrake 1.8.1 Data

Product	Value
AMD Ryzen 7 7700X3D	70.8
AMD Ryzen 7 7800X3D	76
AMD Ryzen 7 7700X	79
Intel Core Ultra 5 250K Plus	101
AMD Ryzen 7 9800X3D	92

The Core Ultra 5 250K Plus demonstrates that its compute performance is peak for a budget-oriented processor, offering more rounded compute levels than the gaming-focused X3D chips.

But if you're after a bit more than just gaming performance from your processor then there are other options you should be considering. Obviously it's a different platform, so if you're just talking about an upgrade to your existing AM5 system then it's not an easy option, but Intel's excellent Core Ultra 5 250K Plus is an outstanding budget processor.

Its eighteen cores of Arrow Lake processing power gives it the edge over all of AMD's eight-core, 16-thread chips when it comes to actual compute power. And it's none too shabby when it comes to the gaming performance in general, either. So, if you wanted something a bit more rounded than a pure-blood gaming chip, the Intel CPU is an outstanding option for just $200.

AMD Ryzen 7 7700X3D thermals

Despite some Zen 4 CPUs being rather toastie when it comes to their cooling demands, the 7700X3D remains pretty chill, largely thanks to that slower clock speed. Because of that it's drawing less power from the PSU, with the chip pulling down almost half that of the straight 7700X during gaming workloads. As we've noted, that lower clock speed doesn't actually have a huge impact on its gaming performance, but does mean that you're only looking at around 61 °C on average in our looped Metro Exodus stress test.

Thermal and power performance

We use three looped runs of Metro Exodus: Enhanced Edition to get a bead on sustained thermal and power draw

Gaming thermalsGaming power

AMD Ryzen 7 7700X3D

61

AMD Ryzen 7 7800X3D

64

AMD Ryzen 7 7700X

69

Intel Core Ultra 5 250K Plus

57

AMD Ryzen 7 9800X3D

51

020406080

Avg CPU temp (°C)

Gaming thermals Data

Product	Value
AMD Ryzen 7 7700X3D	61
AMD Ryzen 7 7800X3D	64
AMD Ryzen 7 7700X	69
Intel Core Ultra 5 250K Plus	57
AMD Ryzen 7 9800X3D	51

AMD Ryzen 7 7700X3D

53

AMD Ryzen 7 7800X3D

69

AMD Ryzen 7 7700X

91

Intel Core Ultra 5 250K Plus

91

AMD Ryzen 7 9800X3D

67

0255075100

Avg package power (W)

Gaming power Data

Product	Value
AMD Ryzen 7 7700X3D	53
AMD Ryzen 7 7800X3D	69
AMD Ryzen 7 7700X	91
Intel Core Ultra 5 250K Plus	91
AMD Ryzen 7 9800X3D	67

The 7700X3D ends up being one of the coolest chips in our test, and also the one that consumes the least power.

Referring back to the Intel option, however, and while that might be a power-sucking beast when it comes to compute-heavy workloads, it is one of the coolest-running of our tested chips here when it comes to gaming performance. You're looking at just 57 °C in our looped Metro Exodus stress test, even while drawing an average 91 W.

AMD Ryzen 7 7700X3D value

Notions of value are a tricky thing to realise for the Ryzen 7 7700X3D. Kudos to AMD for releasing a similar CPU to the 7800X3D yet with a massively lower MSRP, but the realities of retail means that the two processors currently exist at effective price parity. The 7800X3D has been out for a long time now, has a lot of stock still in retail and is regularly discounted.

And while we have noted that higher clock speeds don't necessarily always translate into higher gaming performance, I would have more confidence in the established silicon of the 7800X3D, especially when I'm still uncertain over the provenance of the 7700X3D chips' CCDs.

Are they cores that couldn't hit 7800X3D clock speeds out of the factory and so had to be managed down? Else why not just make more 7800X3Ds? This is my conundrum, and why, if the prices are the same, I would spend the same money on the older, quicker chip.

But the DIY market is not the only market. So while there is price parity here at Amazon for chips purchased on their lonesome, there are other ways to access the CPU. We've seen some excellent 7700X3D chip/mobo/RAM bundles on Newegg that come in at a cheaper price than equivalent 7800X3D options, and I would not be surprised to see full system builds from PC makers coming in undercutting 7800X3D machines.

In that context, the value proposition of the 7700X3D is far more pronounced. The only fly in the AMD ointment here is that Intel's excellent Core Ultra 5 250K Plus is still just $200. That is a bargain price for a chip that is far better at productivity and gets mighty close when it comes to gaming performance, too.