The Best Graphics Cards for 2025
PCMag editors select and review products independently . If you buy through affiliate links, we may earn commissions, which help support our testing .
Gamer or content creator, you've probably given more thought (and cash!) to your graphics card than any other desktop PC component. Though originally designed for gaming, graphics cards are also powerful workhorses for program development, creative work, data analysis, and countless other tasks. At PCMag, we've been testing cards since before the term GPU (which stands for "graphics processing unit") was common parlance. We've tested samples of nearly every card from AMD, Intel, and Nvidia for the last decade-plus. Today, our team tests cards across a host of synthetic tests and AAA games of various genres to gauge raw performance. Our current picks for the best graphics cards for basic 1080p gaming are Nvidia's GeForce RTX 4060 and AMD's Radeon RX 9060 XT, with the RTX 5060 likely to eclipse the 4060 once we are done testing and reviewing it. (We have additional picks for different budgets and resolution targets.) Whether you're buying your first graphics card or upgrading your current one, we'll help you find the right one.
First off, what does a graphics card do? And do you really need one?
If you're looking at any given prebuilt desktop PC on the market, unless it's a gaming-oriented machine, PC makers will de-emphasize the graphics card in favor of promoting CPU, RAM, or storage options. Indeed, sometimes that's for good reason; a low-cost PC may not have a graphics card at all, relying instead on the graphics-acceleration silicon built into its CPU (an "integrated graphics processor," commonly called an "IGP").
A modern graphics solution, whether it's a discrete video card or an IGP, handles the display of 2D and 3D content, drawing the desktop, and decoding and encoding video content in programs and games. All of the discrete video cards on the consumer market are built around large graphics processing chips designed by AMD, Intel, or Nvidia. These processors are referred to as "GPUs," for "graphics processing units," a term that is also often applied, confusingly, to the graphics card itself.
Now, there's nothing inherently wrong with relying on an IGP—most business laptops , inexpensive consumer laptops, and budget-minded desktops have them. If you don't plan to play games or do a lot of content creation work, chances are the IGP will work fine for you, but if you're a gamer or a creator, the right graphics card is crucial.
Even if you do plan to play games, if you only want to play games in your web browser or older games (from, say, five-plus years ago), the integrated graphics may also work just fine for you. If your PC doesn't have an IGP, though, then buying a graphics card is essential to have a functional PC. Some CPUs, notably many AMD Ryzen processors pre-2022, simply don't have on-chip graphics to fall back on. Gamers will also want to buy a graphics card to play most games, as even most games from a decade ago will run far better on a graphics card than an integrated solution.
Last but not least, a graphics card can also be a valuable upgrade to any office PC if that PC is used for a lot of content creation work. Video and image editing programs often support using graphics cards to accelerate work, which can save you a great deal of time and make your work easier.
Graphics cards fall into two distinct classes: consumer cards meant for gaming and light content creation work, and professional workstation cards geared toward scientific computing, intense calculations, and artificial intelligence work. This guide and our reviews focus on the former.
Hardware Limitations: What Graphics Card Is a Good Fit for My PC?
You should have some idea at this point if you still want to get a graphics card or not. But to pick one that you want to buy, you'll need to set a goal for yourself and carefully consider the limitations of your PC. Let's start with the latter: those limitations.
Every PC has a limited amount of room inside its case and a limited amount of power that its power supply can handle. If you don't check the space available in your PC case before buying a card, you run the risk that it won't fit, and you'll be stuck buying a new PC case or having to return your card for another (smaller) one.
If you don't check how much power your PC's power supply can support before buying a GPU, that can be an even bigger deal. If you buy a card that needs more power than your PSU can handle, you could potentially damage your power supply or other parts by repeatedly pushing the power supply past its safe limit. Your system may also seem to work fine most of the time, but then crash periodically whenever the power draw rises too high.
These problems are an inconvenience at the least, and seriously troublesome at worst. Both are easy to avoid if you simply check first what your PC can support before buying. If you are building a new PC from scratch, you'll also want to make sure you buy a power supply and case that can explicitly handle all of your parts.
Nvidia and AMD both outline recommended power supply wattage for each of their graphics card families. These estimates typically overshoot a bit beyond what is actually needed, but you should follow them nonetheless to avoid any issues.
Your system needs to have a PSU that's up to the task of giving a new card enough juice. This is something to be especially wary of if you're putting a high-end video card in a pre-built PC that was equipped before with a low-end card, or with no card at all. Doubly so if it's a budget-minded or business system—these PCs tend to have underpowered or minimally provisioned PSUs.
The two most important factors to be aware of here are the power connectors on your PSU and the maximum wattage the PSU is rated for. The PCI Express (PCIe) slot that the graphics card is plugged into on the motherboard provides some power to the graphics card, but almost all graphics cards (barring a few very low-end ones) require additional power. This is provided by cables, or power leads, that run directly from the power supply to the graphics card.
Traditionally, this has been accomplished by using either six-pin or eight-pin power connections, and these remain common on modern graphics cards and power supplies. They look like this on the card end...
Nvidia, however, introduced its own power connector, which was first used on its GeForce RTX 30-series graphics cards. It has adopted a similar, newer power connector called 12VHPWR for its GeForce RTX 40-series and RTX 50-series graphics cards. The newest connector is pictured below.
Some Nvidia-based cards have this connector on the GPU; others rely on one or more "classic" six-pin and/or eight-pin connections. Pay attention to the specific connections that a graphics card you are considering needs, but you shouldn't have to worry too much, as Nvidia's card-making partners provide adapters, like the one pictured below, for its graphics cards in the box where needed, making plugging them into mainstream power supplies mostly a non-issue. Just make sure you have enough connections on your power supply of the right kind, and that the supply meets the recommended minimum wattage requirements.
Last, if you are upgrading an existing computer, it's important to know that a graphics card can end up bottlenecked and perform slower than expected if the rest of the system it is installed into isn't fast enough to keep up. This mostly applies to the CPU that you pair the graphics card with and, to a far lesser extent, the system's RAM. We can't give you ironclad, simple recommendations here, as faster GPUs will need faster CPUs, and slower graphics cards will work fine with slower processors. (Look to individual card reviews for more detail on that aspect.) We will give more direct advice here on our individual card recommendations, though.
Meet the Card Players: AMD, Intel, and Nvidia
The graphics card market is dominated by AMD and Nvidia, with chip giant Intel entering the fray in late 2022 with its own line of desktop graphics cards sold under the Intel Arc brand.
Intel: The Arc Newcomer
Intel's top-end graphics card at the moment is the Intel Arc B580, followed by the closely related Intel Arc B570. These graphics cards are based on Intel's second-generation graphics architecture, known as "Battlemage," which significantly improved performance over Intel's first-generation "Alchemist" graphics architecture. The Intel Arc B580 and Intel Arc B570 were also released at relatively low prices of $249 and $219, respectively.
The last-gen Intel Arc A770 and Arc A750 aren't easy options to recommend now, as they have been effectively replaced by the Intel Arc B580 and the Intel Arc B570. The more affordable Intel Arc A580 and Arc A380 at the low end of the spectrum may still be worth considering, but only for their exceedingly low price points. Early inconsistencies in performance and early driver issues plagued Intel's first graphics cards, but these issues have mostly been corrected by now.
Being the CPU giant it is, Intel has been producing IGPs as part of its processor chips since the late 1990s and has enjoyed a long history in the industry. But it has a lot of ground to cover before it can catch up with AMD and Nvidia in the dedicated-graphics world. (Graphics cards are often called "dedicated graphics" to distinguish them from integrated graphics on CPUs.) Nonetheless, its cards still might be worth considering under the right circumstances, especially as improved graphics drivers hit the street in the coming months and years.
AMD: Radeon Is in the Running
AMD sells consumer graphics cards as part of its AMD Radeon RX product line and workstation cards under the AMD Radeon Pro and AMD Radeon Instinct brands. The latter two lines are beyond the scope of this article, with Instinct cards designated for professional AI use and Radeon Pro for workstations.
AMD's current generation of consumer graphics cards is the Radeon RX 9000 series, which made its debut with the Radeon RX 9070 XT and RX 9070. The Radeon RX 9070 is overshadowed by the RX 9070 XT, which costs just slightly more for a substantial step up in performance. The new AMD Radeon RX 9060 XT likewise produces highly competitive performance for its price, and it largely makes AMD's outgoing Radeon RX 7000-series graphics cards no longer worth considering. The only real exception to this comes from the AMD Radeon RX 7600, which remains somewhat viable due to its lower price tag. However, even this is questionable due to the 8GB version of the AMD Radeon RX 9060 XT, which we haven't reviewed yet but is close to the RX 7600 in price.
Nvidia: GeForce Is the Dominant Force
Nvidia splits its products in a similar manner, with the company's mainstream consumer graphics products branded as Nvidia GeForce GTX or RTX (GTX cards are older and phasing out), and its workstation cards branded as RTX A Series (formerly Quadro). Nvidia's newest generation of consumer graphics cards is the GeForce RTX 50 series, with the GeForce RTX 5090 being Nvidia's current flagship and the world's single fastest consumer graphics card. Nvidia has pushed out most of its RTX 50-series lineup in 2025, down to the RTX 5060 at this writing; a GeForce RTX 5050 is expected to debut in mid-to-late July. For now, like in previous generations, the older RTX 40 series of graphics cards is still worth considering if you find them on sale for less than their upgraded RTX 50-series counterparts.
What Are 'Board Partners' and 'Reference Designs'?
AMD, Intel, and Nvidia all design the GPUs that bear their names. The chip makers also produce a limited number of full graphics cards themselves and sell directly to consumers. But most graphics cards are created by other companies that are referred to as "board partners."
Board partners may not design and make the actual graphics chips at the core of their cards, but they nonetheless play a pivotal role in the production of graphics cards. They design the physical cards, the power systems, and the thermal solutions that make these cards possible. You'll see a lot of board partners at this point as you shop, including Acer, ASRock, Asus, Biostar, Colorful, Galax, Gigabyte, Inno3D, MSI, Palit, PowerColor, PNY, Sapphire, XFX, and Zotac. Some offer a mix of cards based on AMD's, Intel's, and Nvidia's GPUs. Others limit themselves to cards based on just one GPU maker's chips. (Sapphire, for example, sticks to making AMD-based cards.)
The big three GPU makers often work up what are known as "reference designs" for their video cards, a standardized version of a card built around a given GPU. Sometimes these reference-design cards are sold directly by the GPU maker; more often, though, the chip makers work with the board partners to market and incorporate their GPUs into cards the partners sell. That's why you'll see, for example, GeForce RTX 5080 cards sold by a host of companies that are not actually Nvidia itself.
Depending on the graphics chip in question, these board partners may sell their own self-branded versions of the reference card (adhering to the design and specifications set by the GPU maker), or they will fashion their own custom products, with different cooling-fan designs, slight overclocking done from the factory, or features such as LED mood illumination. Some board partners will do both—that is, sell reference versions of a given GPU, as well as their own, more radical designs.
Though graphics card designs can vary greatly from one company to another, it's important to note that the graphics processors used by these companies are all essentially the same. In other words, one Nvidia GeForce RTX 4080 Super graphics card, for example, will have the same graphics processor and roughly similar performance to all other RTX 4080 Super graphics cards, regardless of which board partner made it.
That's not to say that all Nvidia GeForce RTX 4080 Super graphics cards will be entirely the same, however. Differences in power design, thermal hardware, and clock speeds can result in some performing better than others in different partner cards. Quantifying the exact difference is difficult without detailed benchmark testing, but we can recall differences of up to 12% in the past, and slightly more than that is realistically possible within a given class of GPU.
Graphics Card Basics: Cores, VRAM, and More
Before going further, we need to go over some of the basic parts of a graphics processor, so that you can better gauge how GPUs compare to each other. Without going too deep, graphics processors contain hundreds, if not thousands, of separate processing elements, and there are multiple types of these processing elements, too.
The most common processing element in any graphics processor is what we typically refer to as a shader core, but each company has its own name for these cores. Nvidia calls the shader cores in its GPUs "CUDA cores," while AMD calls the shaders in its GPUs "Streaming Processors." Intel refers to its GPU shaders as "Vector Engines."
Functionally, all of these are similar in that they perform the same basic role inside the GPU, but their internal designs are totally different, resulting in major differences in performance. As a result, you cannot directly compare any of these against each other.
Though you cannot compare CUDA cores against AMD Streaming Processors or Intel Vector Engines, you can roughly compare cards from the same company and in the same product line by comparing the core count. In general, the more shader cores, the better. For example, Nvidia's GeForce RTX 4080 Super has 10,240 CUDA cores, while the GeForce RTX 4070 Ti Super has 8,448 CUDA cores. The GeForce RTX 4080 Super is the faster of the two cards, and the higher core count is part of the reason it is speedier.
You shouldn't rely on comparisons of core count alone when making your decision, though, as it is not the only variable that affects performance. It also shouldn't be relied on when comparing different graphics-card generations, like between the Nvidia GeForce RTX 30 series and 40 series, for example, as many other aspects change that make the comparison by core counts less accurate.
Another important detail to watch out for is the clock speed at which the shaders operate. This is every bit as important as the total number of shader cores, and it has a noticeable impact on performance. Most graphics cards today are sold with two clocks listed. One of these is called the base clock, which is essentially a clock that the GPU can run safely for long periods.
The other clock listed is the GPU boost clock, which is a speed toward which the card will increase so long as it has enough power available and is operating cool enough to do so safely. In practice, graphics cards tend to operate somewhere between these two numbers most of the time.
The last basic part of any graphics card that needs discussion is the graphics card's memory, which is also called "VRAM," for video RAM. Modern graphics cards require a great deal of on-card memory and memory bandwidth to work quickly. The amount of bandwidth needed depends on the card in question, but when it comes to the amount of video RAM available, more is essentially always better.
The amount of RAM on graphics cards increased relatively slowly in the last few decades, but it has recently exploded, as cards went from just 3GB or 4GB of VRAM a few years ago to as much as 24GB or 32GB now at the highest end. The size of games has similarly exploded, resolutions and frame rates have shot up, and textures have gotten more and more complex. At this point, it's hard to argue that any amount of VRAM is too much.
Having more VRAM can always come in handy, but it is worth mentioning that the benefit you gain from large amounts of VRAM differs depending on what resolution you play games at. Most games running at 1080p won't gain much benefit from having more than 4GB of VRAM. If you are gaming at 2K resolution, you'll ideally want to have more than this, though, with 8GB being sufficient for most games.
Gaming at 4K is extremely memory-intensive. Depending on the game and game settings, it can be done, in theory, with cards containing as little as 4GB of VRAM. Most games are going to want a lot more than that, though, and realistically, it's too soon to say if 24GB or 32GB is really enough for all games at 4K.
Advanced Graphics Card Features
In recent years, a number of graphics-card-related features have been introduced aimed at improving your gaming visual experience. These items are quite common, but not quite universal, which is why we opted to break them out here by graphics card families. Each graphics card family typically shares feature parity across the line, which makes these relatively easy to group together.
Ray-Tracing Features
First up is ray tracing. Ray tracing is an advanced graphics technique that's extremely demanding but produces exceptional image quality. Specifically, this is a lighting technique that traces the paths that beams of light take from their in-game light sources as they hit and travel through various on-screen objects. The idea is to make the light look more realistic and create convincing reflections and shadows at the same time.
Ray tracing has now been adopted by all three of the leading GPU makers: AMD, Intel, and Nvidia all have dedicated resources inside their graphics processors to handle ray tracing. To take advantage of this, you'll simply need to buy an AMD Radeon RX 6000 series or later graphics card, an Nvidia GeForce RTX graphics card of any stripe, or an Intel Arc graphics card.
AMD and Nvidia both have older-generation graphics processors that lack ray tracing support still floating around on store shelves, and you may see these while shopping. But they are best avoided at this point, except as extreme budget buys. Instead, you should look for something more modern in one of the product lines mentioned above.
AMD FSR, Intel XeSS, and Nvidia DLSS
You should also be aware of FSR, XeSS, and DLSS. These technologies all work in a similar way, but they are not quite universal. XeSS only works on Intel cards, and DLSS only works with select Nvidia cards. AMD created its FSR technology to be more open, and it will actually work on most video cards today, but this varies somewhat from game to game and from card to card.
Essentially, these tools work by reducing the resolution at which the card renders a game (doing so requires less resources than rendering at a higher resolution), then "upscaling" the results to allow for better frame rates with less loss of quality. For example, imagine you are playing a game at 4K with one of these technologies enabled. Instead of rendering the game at 4K, depending on the settings, the game might be rendered at 2K instead. Upscaling work is then performed by other parts of the graphics card with the assistance of AI-powered algorithms. This may not give you the full image quality of "true" 4K, but it should be close, and it will improve performance, delivering more actual frames per second (fps). You should also see better image quality than you would have had you simply dropped your resolution to 2K.
DLSS comes in a few successive versions, and different games may support different levels of it. One of the newer flavors, DLSS 3, is exclusive to Nvidia's GeForce RTX 40 and 50 series, and it works in a different way. For more information on it, check out our review of the Nvidia GeForce RTX 4090 . DLSS 3 itself has now been joined by DLSS 4 on the latest RTX 50 GPUs; DLSS 4 can use a new so-called "Transformer" AI model and generates a larger number of frames. DLSS 4 can be paired with Multi-Frame Generation (MFG), in which the GPU can take classically rendered frames and generate up to three complementary frames via AI to turbocharge frame rates. (Again, though: All this requires games to be compatible with the tech; also, anti-latency functions complement these features to help reduce perceived latency.) AMD supports similar features with its FSR 3 and FSR 4 technologies, while Intel's XeSS 2 is comparable to DLSS 3 or FSR 3.
Image quality is always reduced, to some extent, if you use one of these technologies, but if you aren't able to run a game smoothly at a given resolution, it's worth trying one of them where you can, as it might give you the best results your hardware can achieve. These technologies are also only supported in a subset of games currently, and you might want to pick a card based on which games support these features.
AMD FreeSync and Nvidia G-Sync: No More Tears
Display refresh-rate technologies, like AMD FreeSync and Nvidia G-Sync, are worth mentioning in this article, but they aren't likely to affect your purchasing decision as much as they might have in years past. Both of these work to match the monitor's refresh rate to the frame rate put out by the graphics card, and both have limitations. (The aim is to enhance in-game smoothness and avoid screen "tearing.") FreeSync is far more common; it's an open standard, and it works with most graphics cards, including many Nvidia cards.
Nvidia's G-Sync is still around today, but it is far less common. A full G-Sync implementation requires monitors to have special internal hardware to support it (which tends to boost the price). It works only with Nvidia graphics cards. The two upper levels of G-Sync, dubbed G-Sync Ultimate and just G-Sync, are certified by Nvidia at a per-monitor level; G-Sync Compatible panels aren't as stringently guaranteed.
We've tested both, and unless you're competing in a CS: GO or Overwatch pro circuit, you might be hard-pressed to see any consistent difference between the two in the latest models. Screen tearing was a more difficult problem to solve back when G-Sync was first introduced, and these days both FreeSync and G-Sync Compatible monitors work well enough that only expert eyes can tell the difference.
AMD CrossFireX and Nvidia SLI: They're Dead, Jim
AMD CrossFireX and Nvidia SLI are both dead technologies today, but they were around for so long that we need to discuss them, at least to let everyone know that you can't really do this kind of thing anymore. CrossFireX and SLI enabled you to connect two or more graphics cards together in one PC to achieve better performance, generally for PC gaming. It was a cool idea, and at times, it worked well, but it was never consistent.
For CrossFireX or SLI to work well, games and drivers needed optimizations to support the technology. If they weren't done, you might get worse performance than if you had just one card. At a certain point, AMD and Nvidia phased out the idea and quit making cards with the hardware bridge connectors and logic to support it. As a result, it's pretty much dead today unless you're trafficking in older cards, though it would still be interesting to see it revived. You're definitely best off nowadays just buying the single best card you can get.
Why Your Monitor Matters: Targeting the Optimal Resolution
Resolution is the horizontal-by-vertical pixel count at which your video card will drive your monitor. This has a huge bearing on which graphics card to buy for gaming. If you are a PC gamer, a big part of what you'll want to consider is the resolution(s) at which a given video card is best suited for gaming.
Nowadays, even low-end cards are able to display everyday programs (word processors and the like) at lofty resolutions like 3,840 by 2,160 pixels (a.k.a., 4K), if your monitor supports it. But for strenuous PC games, those cards will not have nearly the power to drive smooth frame rates at those high resolutions. In games, the video card is what calculates positions, geometry, and lighting, and renders the on-screen image in real-time. Raising the resolution and applying higher graphics settings in a game both have a similar effect and will require increasingly more graphics-card muscle to run games smoothly.
What Resolution Will You Play At?
The three most common resolutions at which today's gamers play are 1080p (1,920 by 1,080 pixels), 1440p (2,560 by 1,440 pixels), and 2160p or 4K (3,840 by 2,160 pixels). Generally speaking, you'll want to choose a card suited for your monitor's native resolution. (The "native" resolution is the highest resolution supported by the panel, at which it looks the best.)
You'll also see ultra-wide-screen monitors with in-between resolutions (3,440 by 1,440 pixels is a common one); you can gauge these versus 1080p, 1440p, and 2160p by calculating the raw number of pixels for each (multiply the vertical number by the horizontal one) and seeing how the wide resolution fits in relative to the most common three. (See our targeted roundups of the best graphics cards for 1080p play and the best graphics cards for 4K gaming .)
Why does this matter? Well, a gaming PC needs to be balanced for you to get the most out of your parts. If you have a high-end graphics card and a low-end processor, you might find your graphics card constrained; you get poor performance because the CPU simply can't keep up. The same can also happen in reverse. In general, gaming at higher resolutions is far more stressful on the graphics card than on the CPU, and it also requires far more VRAM. Gaming at lower resolutions, conversely, tends to be harder on the system's CPU.
The reason for this is that the amount of work required by the processor to create each frame doesn't change all that much as you change the resolution. This is again the opposite for the graphics card, which has to do roughly four times the amount of work to render a 4K frame as compared with a 1080p one. When you game at higher resolutions, then, the graphics card is typically what hits its limits first. At lower resolutions, where the graphics card might be able to output twice as many frames, however, this will double the work required by the processor to keep up, increasing the strain there.
Ideally, you'd want to pick parts that are both capable of running at nearly 100% usage when used in conjunction with each other. Doing so definitively is difficult, though, as the work required varies greatly from one resolution to the next and from game to game. Instead, it's typically better to opt for a processor that won't limit your graphics card at the resolution you intend to play at the most, as you can always lower the frame rate on your system with V-Sync or a similar setting to help avoid or negate the effects of the processor not keeping up. (For more details about which processor would be best for gaming within your budget, check out our best CPUs page .)
Now, of course, if your graphics card isn't fast enough, you can always dial down the detail levels for a game to make it run better at a higher-than-recommended resolution, or dial back the resolution itself. The highest-end cards are meant for playing at 4K, or at very high refresh rates at 1080p or 1440p; you don't have to spend $1,000 or even half that to have an enjoyable gaming experience at any resolution, if you are willing to adjust settings.
Plenty of current-generation midrange GPUs can power 1440p displays at their peak, and 4K gaming isn't even out of the question for a midrange GPU...though most gamers aren't looking to play at either resolution, if the Steam Hardware Survey is any indication. (It saw less than 5% of users playing at resolutions higher than 1440p at this writing.)
High-Refresh Monitors: Why High-End GPUs Matter
The other trend in gaming driving GPU demand in recent years? High-refresh-rate gaming monitors. For ages, 60Hz (or 60 screen redraws a second) was the limit for most PC monitors, but that was before esports hit their stride. Panels focused on esports and high-refresh gaming may support up to 144Hz, 240Hz, or even 360Hz for ultrasmooth gameplay, with a few extreme 500Hz-plus panels also out there. If your video card can consistently push frames more than 60fps in a given game, a high-refresh monitor lets you see those formerly "wasted" frames in the form of smoother game motion.
Esports has boosted the demand in recent years for high-refresh monitors that can keep hopefuls playing at their peak. And while 1080p is still overwhelmingly the preferred resolution for competitive players, the 1440p bracket of graphical resolutions (played in either 16:9 aspect ratio at 2,560 by 1,440 pixels, or in 21:9 at 3,440 by 1,440) is growing faster than ever. It all depends on the way you prefer to play, as well as on which games you play.
Most casual gamers won't care about extreme refresh rates, but the difference is marked if you play fast-action titles, and competitive esports hounds find the fluidity of a high refresh rate a competitive advantage. (See our picks for the best gaming monitors , including high-refresh models.) In short, buying a powerful video card that pushes high frame rates beyond 60fps can be a boon nowadays, even for playing games at a "pedestrian" resolution like 1080p if paired with a high-refresh monitor.
What Connections Should My Graphics Card Have?
Three kinds of ports are common on the rear edge of a current graphics card: DVI, HDMI, and DisplayPort. Some monitors still use DVI, but it's the oldest of the three standards and no longer appears on mainstream and high-end cards, just the occasional low-end card meant for use as a simple display adapter.
Most cards have several DisplayPorts (often three) and one HDMI port. When it comes to HDMI versus DisplayPort, note some differences. First, if you plan on using a 4K display, now or in the future, your card needs to at least support HDMI 2.0a or DisplayPort 1.2/1.2a. It's fine if the GPU supports anything above those labels, like HDMI 2.0b or DisplayPort 1.4, but that's the minimum you'll want for smooth 4K playback or gaming. Recent-generation cards from all the makers will be fine on this score.
HDMI 2.1 is a newer cable spec you may see mentioned, which raises the old bandwidth limits from 18Gbps (in HDMI 2.0) to 48Gbps (in HDMI 2.1). The upgrade enables eventual 8K resolution to display at a refresh rate up to 60Hz, with 4K supported up to 120Hz. At this point, it's more about future-proofing than much else and accommodating the latest very high-end/high-refresh 4K monitors.
Note: Some of the cards from Nvidia's GeForce RTX "Turing" line (the RTX 20 Series, which is getting old now) that you may still see around employ a port called VirtualLink. This port looks like (and can serve as) a USB Type-C port that also supports DisplayPort over USB-C. What the port was really designed for, though, is attaching future-gen virtual reality (VR) headsets. It never took off and has disappeared from the latest cards.
GPU Budgeting: How Much to Spend on a Graphics Card?
Everything that we've written in this article so far has been to give you a basic understanding of the graphics card market (and the hardware) to help you make an informed decision. We have more specific advice in our top picks around which graphics card is best at each resolution and for different people. But ultimately, price is the biggest and most important factor.
No matter what games you want to play or what resolution you intend to play at, you should buy the best graphics card that fits your budget for your new PC or your upgrade. To be clear, we aren't suggesting everyone go buy an Nvidia GeForce RTX 5090. However, it's better to pay out more for a graphics card that might be a bit faster than you need...than it is to buy one you think will skate by your requirements, and then be disappointed.
A key thing to keep in mind: Games are constantly becoming more graphically intensive, baseline resolutions keep increasing, and faster monitors that support ever-higher frame rates keep coming out. Unless you don't game much, it's not going to be long before you get a new monitor or find a new game that's more demanding on your PC.
When that happens, you may well want to upgrade your graphics card again, no matter what graphics card you buy now. If you opt for the best card that you can afford, however, you put off that day that much longer, and save money long-term. And, as an added benefit, you'll likely be able to run the games you do play now with slightly better settings.
To be perfectly clear, no graphics card is, by default, always a bad choice at a given resolution; everything depends on a mix of factors. Take Nvidia's GeForce RTX 5090, which is the single fastest graphics card we have ever tested. You might think it is overkill for 1080p, and it will be, for most people. But if you have a high-refresh-rate 500Hz/1080p monitor, it actually might be the best option for you in certain games. We haven't tested any graphics cards that can unquestionably exceed 60fps in all games at all graphics settings, but the RTX 5090 does the best and will let you obtain a higher refresh rate than any other option currently available.
Similarly, you might think some more budget-friendly cards, such as the Intel Arc B580, aren't suitable for 4K gaming. But in our review, it was able to play several games well enough at 4K with maxed-out graphics settings. You'd want to turn down the graphical detail a bit in some games to maximize frame rates, but the point is, if that's the best card you can afford on your budget, and you are dead-set on playing at 4K, you still can.
What Is the Highest-Rated Graphics Card?
We evaluate graphics cards based on performance, pricing, and value for money. We haven't had a graphics card with a perfect 5-star score since 2020's GeForce RTX 3070 Founders Edition . (That card's still pretty good if you can find one, but it's been supplanted by two generations now.)
Our current highest-rated graphics cards, at 4.5 stars, are the Intel Arc B580 Limited Edition, the Sapphire Pulse AMD Radeon RX 9070 XT, the Nvidia GeForce RTX 4060, and the Nvidia GeForce RTX 5080 Founders Edition.
Ready to Buy the Best GPU for You?
You won't find a single right answer to the question "What's the best graphics card for me?" It's all about your budget, your monitor, your existing system, and what games you play. Think carefully about what you want out of your system and what you can afford to spend on a graphics card. Make absolutely sure it's going to fit in your case and that your power supply, CPU, and other supporting parts are up to the job, and then buy the best graphics card you can manage. Don't go over your planned budget, as you can always buy a better graphics card down the road, but shop within reason, because the best card you can afford now is likely the best for you. You're now ready to look at our parts picks, all lined up, and buy with confidence.
