Best graphics card 2025

What is the best graphics card? Finding a new gaming GPU can mean wading through a swamp of often baffling numbers in specs lists, as well as weighing up the pros and cons of the different feature sets from Nvidia, AMD, and Intel. Fear not, though, as we’ve done all the testing and comparisons for you so that our best graphics card guide will steer you towards the right card for your needs without the jargon.

If you just want the best, the Nvidia GeForce RTX 5090 is the one. Yes, it costs at least $2,000 but it is by far the fastest GPU of all and has access to Nvidia’s latest DLSS 4 multi-frame gen tech, which you can learn more about in the how to choose a graphics card of this guide. If, like most of us, cost is actually a consideration for you, though, the new AMD Radeon RX 9070 XT is the card to go for, as it offers fantastic performance for its $599 MSRP. Read on to find plenty of other options to suit other budgets, and also check out this guide’s FAQ section that answers all your graphics card questions.

AMD Radeon RX 9070 XT

Best graphics card for most

Today’s best deals

AMD Radeon RX 9070 XT specifications:

GPU architecture	AMD RDNA 4
Stream processors	4,096
AI cores	128
RT cores	64
Game clock	2,400MHz (2.4GHz)
Boost clock	2,970MHz (2.97GHz)
VRAM	16GB GDDR6
Memory Bandwidth	644.6GB/s
TDP	304W

Pros

• Incredible gaming frame rates
• 16GB of VRAM
• Great value for money

Cons

• Draws a lot of power
• Not many games support FSR 4
• No equivalent of multi frame gen

AMD has stormed back into the PC gaming GPU arena with all guns blazing, and the new Radeon RX 9070 XT is the company’s most competitive graphics card for a number of years.

It costs just $50 more than the Nvidia RTX 5070, and it’s significantly more powerful across the board, while also having a full 16GB of VRAM for future proofing. What’s more, AMD has massively caught up with the competition when it comes to ray tracing and AI. This new card supports AMD’s FSR 4 resolution upscaling tech, which looks massively better than FSR 3, and its GPUs can keep up with Nvidia in ray tracing now, while also handling path tracing.

There are only a couple of caveats about this new GPU. Firstly, it draws a lot of power, so make sure you have at least an 850W PSU in your rig. AMD also doesn’t have an equivalent of Nvidia’s multi frame gen tech yet either, where a GPU can use AI to insert multiple frames between genuinely rendered ones, smoothing out your frame rate. However, in terms of raw gaming power, the 9070 XT is absolutely the best buy in this price sector right now.

Read our full AMD Radeon RX 9070 XT review.

Nvidia GeForce RTX 5090

Best 4K graphics card

Today’s best deals

Nvidia GeForce RTX 5090 specifications:

CUDA cores	21,760
Tensor cores	170 (4th-gen)
RT cores	680 (5th-gen)
Boost clock	2,407MHz
Base clock	2,017MHz
VRAM	32GB 28Gbps GDDR7
Memory bandwidth	1,792GB/s
TGP	575W

Pros

• Multi Frame Gen is amazing
• Fastest gaming GPU ever
• Great two-slot cooler

Cons

• Astronomical price
• Scarce supply
• Not all games support Multi Frame Gen

Nvidia has gone all out with its latest flagship, which brings the full magic of DLSS 4 to 4K gaming with Multi Frame Gen. This clever new tech uses AI to insert up to three extra frames between each pair of rendered frames, and it massively increases the frame rate.

In fact, in our tests, this card can even run Cyberpunk 2077 at 4K with full path tracing enabled and get an average of 207fps, a phenomenal result at these settings. Comparatively, the RTX 4090 can only average 82fps in this game at its maximum frame gen settings. What’s more, thanks to the new transformer model Nvidia has introduced with DLSS 4, your games look much sharper if you enable the Super Resolution upscaling tech.

Meanwhile, Nvidia has also managed to cram a load more CUDA cores into this card’s colossal GPU, giving it a decent step up in performance over the 4090 across the board. This performance benefit is smaller than we’ve seen in previous generational shifts. For example, the RTX 5090 averages 146fps in Call of Duty Black Ops 6 at 4K, compared to 120fps on the RTX 4090, but that’s still a performance boost.

It’s not all plain sailing, which is why the RTX 5090’s review score is only a 6 right now. The $1,999 MSRP is already sky-high, and the actual prices of board partner cards are much higher. Plus it can be hard to find RTX 5090 stock right now, unless you want to pay scalper prices. The power draw is also extreme, and you’ll need at least a 1,000W PSU to feed this beast of a graphics card.

If you want the absolute fastest GPU you can buy, though, and you’re prepared to pay the price, make no mistake – this is the card you want.

Read our full Nvidia GeForce RTX 5090 review.

Nvidia GeForce RTX 5080

Best graphics card under $1,000

Today’s best deals

Nvidia GeForce RTX 5080 specifications:

CUDA cores	10,752
RT cores	336 (5th-gen)
Tensor cores	84 (4th-gen)
Base clock	2,300MHz
Boost clock	2,620MHz
VRAM	16GB 32Gbps GDDR7
Memory bandwidth	960GB/s
TBP	360W

Pros

• Much cheaper than RTX 5090
• Decent 4K gaming performance
• DLSS 4 with Multi Frame Gen is awesome

Cons

• Hard to find in stock at MSRP
• Not much faster than 4080 Super without DLSS 4
• Not all games support Multi Frame Gen

If you can’t afford the two grand for the RTX 5090 (and that’s quite understandable), but you want to play games at 4K with decent settings, then this $999 GPU from Nvidia is the best in its class. Like the 5090, it also supports Nvidia’s latest Multi Frame Gen tech, meaning it can use AI to insert two or three or more frames between each pair that your GPU renders, enabling you to max out the eye candy in your games and still get smooth frame rates.

We tested the new tech with Cyberpunk 2077, maxed out with path tracing so it looks incredible, and the RTX 5080 averaged 128fps. Drop down to 1440p and you’re looking at 228fps, which is amazing in a game that’s this demanding. As a point of comparison, even the RTX 4090 only averages 148fps in this game at 1440p with its maximum level of frame gen enabled.

Not all games support this new tech, of course, but Nvidia has announced 75 Multi Frame Gen games at launch, and if your game already supports standard frame gen, you can force it to use Multi Frame Gen using the DLSS Override feature in the Nvidia App.

Without Multi Frame Gen, the RTX 5080 is only a little bit faster than the RTX 4080 and 4080 Super, which is a little disappointing, but at least the new GPU maintains the $999 price of the latter, rather than the obscene $1,199 launch price of the original RTX 4080.

Of course, that latter point is largely moot if you can’t find a card at MSRP, and the current short supply of the RTX 5080 means it’s hard to buy one right now. You can check out our where to buy 5080 page for the latest stock updates in your area, but if you can get this GPU for $999, it’s absolutely worth it.

Read our full Nvidia GeForce RTX 5080 review.

AMD Radeon RX 9070

Best sub-$550 graphics card

Today’s best deals

AMD Radeon RX 9070 specifications:

GPU architecture	AMD RDNA 4
Stream processors	3,584
AI cores	112
RT cores	56
Game clock	2.07GHz
Boost clock	2.52GHz
VRAM	16GB GDDR6 20Gbps
Memory bandwidth	644.6GB/s
TDP	220W

Pros

• Solid frame rates
• Loads of VRAM
• Power draw is reasonable

Cons

• Might as well buy the 9070 XT at MSRP
• No multi frame gen
• Limited availability

When you only need to spend $50 more to get the fantastic 9070 XT, the 9070 gets a little bit lost in this crowded market, but if you genuinely can’t afford the extra money for the 9070 XT, or if you simply can’t find the latter in stock, the Radeon RX 9070 is a solid option.

Not only does it have more VRAM than the competing Nvidia GeForce RTX 5070, giving it some more future proofing, but it’s also more powerful in most of our game tests. In fact, in Call of Duty Black Ops 6, it’s even faster than the RTX 4080.

Unlike previous AMD GPUs, the 9070 can also cope well with ray tracing, where it out-performs the RTX 5070 in F1 24, and isn’t far off the pace in Cyberpunk 2077. The Nvidia GPU is quicker at path tracing, and also has the benefit of multi frame gen, but the 9070 is generally the more powerful option.

Not only that, but it’s also more power-efficient than the 9070 XT, with our test system drawing 395W from the mains with the 9070 installed, compared to a massive 487W with the 9070 XT.

Read our full AMD Radeon RX 9070 review.

AMD Radeon RX 7800 XT

Best graphics card under $500

Today’s best deals

AMD Radeon RX 7800 XT specifications:

GPU die	Navi 32 XT
Stream processors	3,840
RT cores	60 (Second Gen.)
Base clock	1,624MHz (1.62GHz)
Game clock	2,124MHz (2.12GHz)
Boost clock	2,430MHz (2.43GHz)
VRAM	16GB GDDR6
Memory bandwidth	624GB/s
TBP	263W

Pros

• Unparalleled memory bandwidth for $500
• Smashes the GeForce RTX 4070 in rasterized scenarios
• Better value than the Radeon RX 7700 XT

Cons

• Large ray tracing performance gaps vs the RTX 4070
• Still no sign of FSR 3 frame generation
• Some generational downgrades

While the Radeon RX 9070 cards are grabbing all the headlines right now, they’re notoriously hard to find in stock, particularly at MSRP. That’s where the AMD Radeon RX 7800 XT makes for a good backup plan if you really need a mid-range GPU, especially now that the RTX 4070 is hard to find too.

For those who aren’t too fussed about ray tracing, this is the graphics card to get. With 16GB of VRAM attached to a 256-bit memory interface, you get a decent memory configuration for the money. This uniquely positions the 7800 XT as the only card at the $500 mark that’s capable of 1080p, 1440p, and even some 4K gaming.

It’s the more sensible buy vs the AMD Radeon RX 7700 XT, given that there’s only $50 separating the two. In our testing, the RX 7800 XT actually beat the more expensive RTX 4070 at every resolution up to 4K with ray tracing disabled, sometimes by as much as 17% on average. However, as soon as ray tracing effects enter the frame, it falls behind and has no answer for DLSS 3 either.

Read our AMD Radeon RX 7800 XT review.

AMD Radeon RX 7600

Best cheap graphics card

Today’s best deals

AMD Radeon RX 7600 specifications:

Stream processors	2,048
RT cores	32 (Second Gen.)
Base clock	1,720MHz (1.72GHz)
Game clock	2,250MHz (2.25GHz)
Boost clock	2,655MHz (2.66GHz)
VRAM	8GB GDDR6
Memory bandwidth	288GB/s
TBP	165W

Pros

• Less than $300
• Surprisingly good rasterization performance
• Decent TBP

Cons

• 8GB of VRAM
• Subpar ray tracing performance
• No answer to DLSS Frame Generation

If your budget can’t stretch to a midrange graphics card, the AMD Radeon RX 7600 is the way to go. There isn’t a perfect GPU in the budget space right now, but the Radeon RX 7600 is the least compromised option for the money.

Both the RX 7600 and Nvidia GeForce RTX 4060 come with just 8GB of VRAM. This capacity is fine for esports titles at 1080p but can create problems when higher-resolution textures are enabled. Naturally, 1440p and 4K are off the table for this GPU.

Given its low price, we’re surprised how perky this pixel pusher is in the right circumstances. Frame rates are impressive at 1080p with ray tracing disabled, largely beating the GeForce RTX 4060, although we don’t recommend turning on ray tracing effects in demanding games with this graphics card.

Read our AMD Radeon RX 7600 review.

How to choose a graphics card

The first step to finding a good graphics card choice for you is the same as every other component: budget. The Nvidia RTX 5090 is only the fastest option for your system if you can afford it, and knowing how much you can spend on your new GPU will help narrow down your options.

Next up, consider what sort of games you play and what monitor you have. If you mainly play indie games, older titles, or play competitive games at a more casual level, none of these options put a huge demand on your GPU, so you might be able to get away with buying a more entry level GPU such as the AMD Radeon RX 7600. It brings support for many modern features and still has the grunt to power AAA games at 1080p resolutions at medium to high settings but costs well under $300.

Likewise, if your monitor only has a resolution of 1080p and a refresh rate of 120Hz or less, there’s very little point in spending big on a GPU. If you do, you’ll have the horsepower ready to run at higher resolutions and faster frame rates but only an upgrade to one of the best gaming monitors will truly unlock that potential.

Meanwhile, if you have what many consider to be the goldilocks screen size and resolution – a 27-inch, 1440p display – with a 144Hz-240Hz refresh rate, that’s when mid-range GPUs such as the AMD Radeon RX 9070 and RX 9070 XT come in to play. For around $600 these will provide blazing fast frame rates at 1080p and 1440p in most games while bringing steady 60-120fps frame rates and stunning visuals in many AAA games with all the settings cranked up.

If you have a 4K monitor or a super-fast 360Hz+ gaming monitor – or one that can do both, such as you can read about in our LG Ultragear 32GS95UE review – that’s when you’ll want to consider moving up to truly powerhouse GPUs such as the RTX 5080 or all-conquering RTX 5090.

With those core factors taken into consideration, you can now move on to thinking about the finer differences between AMD, Intel, and Nvidia GPUs. All three companies largely offer very similar core features, with all supporting AI acceleration and ray tracing as well as normal game rendering. All also include a type of AI-enhanced upscaling and frame generation technology. AMD has FSR, Intel has XeSS, and Nvidia has DLSS.

On balance, Nvidia is leading the charge on features, thanks mainly to the image quality of its DLSS upscaling and its unique feature called multi-frame generation or MFG. MFG lets a game render one frame in a normal manner then uses AI to insert two or even three extra frames. AMD also has a frame generation technology but it can only insert one frame. Our RTX 5090 review showed just how impressive MFG can be.

While MFG is impressive, and Nvidia’s software technology lead in general isn’t to be dismissed, we’ve seen with the likes of our AMD Radeon RX 9070 XT review that ultimately value trumps all for most buyers and right now AMD holds an advantage here.

FAQs

What is a graphics card?

Graphics cards enable your PC to display images of any type and they also power through the incredibly complex calculations required to render the 3D scenes in games.

Most desktop work and 2D games put a low demand on your graphics card but when games display 3D worlds, the calculation demands increase considerably. As you then move to the stunning, realistic worlds of modern AAA games such as Black Myth Wukong or Kingdom Come: Deliverance 2, a graphics card is performance billions of calculations a second to keep up with demand.

The more powerful modern graphics cards also support ray tracing technology, which is even more demanding than conventional 3D rendering. That’s at least part of the reason we’ve seen the rise in AI to process information as quickly as possible and make your games visually stunning while reducing the overall demand on your GPU.

How do you upgrade your graphics card?

Unless your PC has a custom water-cooling loop that cools your GPU, upgrading your graphics card is one of the easiest and biggest-impact upgrades you can make to your gaming PC. They literally just slot in and out of your PC with just a couple of screws and a clip on the motherboard’s PCIe slot holding them in.

There are really only two factors to worry about, other than deciding which brand you want and how much to spend. The first is whether you have a powerful enough power supply with enough PCIe cables for your graphics card. If you’re upgrading from one older mid-range card to another newer one, chances are you’ll be fine. However, if you’re stepping up from an RTX 2060 to an RTX 5090, you’ll almost certainly need an upgrade to one of the best power supplies to run your new GPU.

The other factor is the length and overall size of the card. Many high-end graphics cards are huge so you’ll need to check the dimensions of the card you’re looking to buy to see if fits in your case. For a complete guide on the process of upgrading your graphics card, check our how to upgrade a GPU guide.

What is VRAM?

VRAM is an acronym for ‘video random access memory’ and refers to the built-in memory used by graphics cards. Its purpose is to provide your GPU with a separate, high-speed pool of memory that it can use to store data.

More VRAM at your disposal is always a positive as it makes your system less reliant on standard RAM to plug any gaps should your graphics card’s capacity become saturated, thereby reducing the chance of stutters. However, capacity is only part of the story, with the speed of your GPU’s memory and its bus width influencing memory bandwidth.

Most modern graphics cards now come with at least 8GB of VRAM which is usually enough for 1080p gaming, with some exceptions. Higher-quality effects and textures demand higher pools of video memory, though, so we’d suggest opting for a 12GB option if you can afford it.

For those of you gaming at 1440p, 16GB is ideal, but 12GB will suit you for the most part. Likewise, 20GB is recommended for 4K, but 16GB will serve.

What is ray tracing?

Ray tracing refers to real-time light tracing to produce more realistic and higher-quality lighting effects than typical rasterized techniques. It has several use cases, including improved shadows, ambient occlusion, reflections, and global illumination.

The first ray-tracing-capable graphics cards arrived in 2018, with the Nvidia GeForce RTX 20 series. AMD joined in on the fun with its Radeon RX 6000 series in 2020, with Intel following with its Arc discrete GPUs in 2022.

Toggling on ray tracing options greatly increases GPU load and can dramatically impact frame rates. As such, it’s typically wise to use them in conjunction with performance-enhancing tools such as Nvidia DLSS, AMD FSR, or Intel XeSS.

What is rasterization?

Rasterization is the process through which 3D models are converted into the pixels you see on the screens of the best gaming monitors. The better a graphics card is at doing this, the smoother your experience of playing your favorite games – when done well, games will look fantastic, when done poorly, games can look blurry and washed out.

Is a GTX or RTX graphics card better?

GeForce RTX graphics cards offer a more feature-rich experience than those sporting a GTX badge and provide superior performance. Comparing GPUs of the same class (e.g. a GTX 1080 against an RTX 2080) will always favor the newer RTX model.

The main distinctions between GTX and RTX graphics cards are ray tracing and DLSS support. Every RTX GPU is capable of using DLSS Super Resolution to boost frame rates via upscaling, in addition to improving visuals through ray-traced effects. Meanwhile, GTX pixel pushers don’t have access to either.

As such, we don’t recommend picking up a GTX graphics card if it can be helped as RTX options provide better value.

Does ray tracing work on all games?

Real-time ray tracing has only been a part of games since 2018, meaning that every game released before then won’t feature the technology unless it’s been patched in since then. However, it’s not a given that every game release after that time will feature ray-traced effects.

Many big games, such as Fortnite and Cyberpunk 2077 offer ray-tracing options, and the number of examples continues to grow. With the advent of RTX Remix, it won’t be long before many classics join newer releases in basking in the light of ray tracing.