Top 5 Best Graphics Cards of 2025

As a gamer or content creator, you’ve probably spent more time and money on your graphics card than any other desktop PC component. Graphics cards, while originally designed for gaming, are also effective workhorses for program development, creative work, data analysis, and a variety of other tasks. gadgetex has been testing cards since before the term GPU (which stands for “graphics processing unit”) became widely used. We’ve tested nearly every card from AMD, Intel, and Nvidia over the last decade.

Today, our team evaluates cards using a variety of synthetic tests and AAA games from various genres to determine raw performance. The current best graphics cards for 1080p gaming are AMD’s Radeon RX 7600 and Nvidia’s GeForce RTX 4060. (We have additional options for various budgets and resolution targets.) Whether you’re purchasing your first graphics card or upgrading an existing one, we’ll help you find the right one.

QUICK LIST

Last update on 2025-03-09 / Affiliate links/images from Amazon.

Top 5 Best Graphics Cards of 2025

1.AMD Radeon RX 7600

Last update on 2025-03-09 / Affiliate links/images from Amazon.

The calm before the storm: AMD’s Radeon team had been surprisingly quiet on the graphics card front in the six months leading up to the product launch. Following the release of its high-end AMD Radeon RX 7900 XTX and AMD Radeon RX 7900 XT, we expected more cards in the near future, most likely in the RX 7800 range as a step down. That hasn’t happened, and we’re only now getting a look at AMD’s next Radeon 7000 Series entry, the midrange $269 AMD Radeon RX 7600.

At this price, the Radeon RX 7600 is a far cry from AMD’s flagship Radeon RX 7900 XTX, which AMD describes as a high-end 1080p gaming solution. The Radeon RX 7600 is up against stiff competition in this market segment. However, for the time being, its combination of excellent 1080p performance and a reasonable price makes it arguably the best modern card for 1080p gaming, earning it our Editors’ Choice.

When AMD first introduced the Radeon RX 7900 XTX and RX 7900 XT, we discussed the new RDNA 3 architecture in depth, as well as the benefits of the new chiplet design used in the card’s construction. We recommend returning to those articles (linked above) for more information on the RDNA 3 architecture. However, the Radeon RX 7600 differs from the flagship in more ways than we would expect.

Chiplet designs are especially useful for semiconductors with a large physical footprint. Breaking them down into smaller modules for use in a chiplet design improves yields and reduces waste, but the smaller the chip, the less effective this is. There are also performance issues with chiplet designs, which AMD has worked hard to address with its Radeon RX 7900 cards and Ryzen processors.

The Navi 33 GPU die for the Radeon RX 7600 is only a fraction of the size of what’s inside of the Radeon RX 7900 XTX. Instead of multiple chips with a combined area of 552 square millimeters, the Radeon RX 7600’s GPU die only measures 204 square millimeters, less than half the size. As a result, AMD didn’t see sufficient benefit to warrant making the Radeon RX 7600 into a chiplet design.

Overall, the Radeon RX 7600 has roughly one-third the compute units available as the Radeon RX 7900 XTX. It has precisely one-third of all hardware resources, including 2,048 stream processors, 128 texture mapping units (TMUs), 64 raster operation units (ROPs), 64 AI accelerators, and 32 Ray Accelerators. Even the memory system was reduced to one-third of what the Radeon RX 7900 XTX has, leaving the Radeon RX 7600 with a 128-bit memory interface and 8GB of GDDR6 RAM. We should also mention that the chip includes 32MB of AMD Infinity Cache.

The only thing that wasn’t reduced by one-third was the card’s clock speeds, which are actually slightly higher in one respect. The Radeon RX 7900 XTX has a slightly higher base clock of 2,300MHz compared to the Radeon RX 7600’s 2,250MHz. However, the more important boost clock on the RX 7900 XTX is limited to 2,500MHz, whereas the boost clock on the Radeon RX 7600 can reach up to 2,625MHz.

The Radeon RX 7600’s PCI Express interface is also reduced by half, resulting in a PCIe 4.0 x8 connection to the host system. While most cards nowadays come with a full PCIe 4.0 x16 connection, a PCIe 4.0 x8 connection should suffice for a card like the Radeon RX 7600.

2.Nvidia GeForce RTX 4060

Last update on 2025-03-09 / Affiliate links/images from Amazon.

When the Nvidia GeForce RTX 4060 Ti was released at the end of May 2023, it didn’t make much of an impact—it was the video-card equivalent. Its position as the best graphics card for 1080p gaming didn’t feel right at that price. Smooth play at 1080p resolution has long been a goal for many gamers, but it is now more of a starting point than an end goal. Paying $399 for fast performance at the lowest resolution anyone cares about today simply isn’t logical. AMD’s latest top graphics card for 1080p gaming, the AMD Radeon RX 7600, which did not help Nvidia’s case for the RTX 4060 Ti either.

GeForce RTX 4060 is Nvidia’s opportunity to retaliate against its lower-cost Radeon competitor. Nvidia’s GeForce RTX 4060 (without the “Ti”) has a much lower price than its bigger brother. We tested the Asus Dual Nvidia GeForce RTX 4060 OC Edition card to see what to expect from the new RTX 4060 GPU series. Off the bat, performance is noticeably slower than the RTX 4060 Ti, and as tech enthusiasts, we can’t help but want more from new components. Still, it’s fast enough to make the RTX 4060, based on this Asus model, the best 1080p graphics card on the market today, earning our Editors’ Choice award in the budget 1080p category.

The GeForce RTX 4060, like all other Nvidia GeForce RTX 40 series graphics cards, is built using a customized 4nm TSMC manufacturing process and Nvidia’s “Ada Lovelace” microarchitecture. If you’re not familiar with the improvements built into this new architecture, read my review of the GeForce RTX 4090 for more information. However, we will briefly review the architecture’s highlights here.

Nvidia’s Ada Lovelace architecture has a number of significant advantages, including the ability to support higher clock speeds thanks to a new manufacturing process. Ray-tracing performance has improved significantly with Ada Lovelace graphics cards, which now include a third-generation ray-tracing core that is faster and more effective than Nvidia’s previous designs. The company also increased the amount of cache on its graphics cards, which I’ll go over more later.

The RTX 40 Series graphics cards’ final and possibly most significant improvement is their support for Nvidia’s DLSS 3 technology. This technology generates new frames by extrapolating them between two existing frames. This can increase frame rates if the graphics card has excess processing power.

Nvidia’s GeForce RTX 4060 is powered by an AD107 GPU core made up of three graphics processing clusters (GPCs) each with eight streaming multiprocessors (SMs), for a total of 24 SMs. Each SM contains 128 CUDA cores, all of which are active inside the GeForce RTX 4060, for a total of 3,072 CUDA cores.

3.AMD Radeon RX 7700 XT

Last update on 2025-03-09 / Affiliate links/images from Amazon.

AMD’s Radeon RX 7700 XT completes the GPU underdog’s RX 7000 series line of battle, designed to compete with the Nvidia GeForce RTX 4060 Ti. AMD isn’t selling an AMD-branded reference version of this graphics card, but we tested an XFX iteration of this GPU, the Speedster Qick 319 RX 7700 XT Black Edition, which costs, to see how the RX 7700 XT performs. And we were left very impressed. Though the RX 7700 XT’s MSRP of is lower than that of the GeForce RTX 4060 Ti, the performance more than compensates for the higher price.

The Radeon RX 7700 XT is based on AMD’s Navi 32 GPU die, but with some resources disabled. This is a chiplet design with a Graphics Chip Die (GCD) and four smaller Memory Cache Dies (MCDs) arranged around it. Though the Radeon RX 7700 XT includes all of these resources, it has fewer of each than the RX 7800 XT.

In total, the Radeon RX 7700 XT has 54 of the 60 possible compute units active in the GCD, resulting in 3,456 streaming processors, 96 ROPs, 54 Ray Accelerators, 108 AI Accelerators, and 216 TMUs. One of the four MCDs is also disabled, reducing the total L2 cache pull from 64MB to 48MB. It also reduces the memory interface width from 256 bits to 192 bits. The total amount of memory available to the card decreases as well, with 12GB of GDDR6 memory on board.

Despite these changes, the Radeon RX 7700 XT remains very similar to the RX 7800 XT, which also uses the Navi 32 die but has no resources disabled. The dropped MCD is by far the most noticeable reduction, reducing memory bandwidth by 25% while only reducing total computational resources by 10%. Higher clock speeds on the RX 7700 XT help to compensate for this, and as a result, the majority of the performance difference between the RX 7700 XT and RX 7800 XT in tests will be due to changes in the memory subsystem.

Before we get into that, let’s take a quick look at the card we tested for this review. As previously stated, AMD did not create a reference version of the Radeon RX 7700 XT; instead, we received an XFX Speedster Qick 319 AMD Radeon RX 7700 XT Black Edition graphics card to represent the platform. This card should demonstrate the RX 7700 XT’s capabilities, as it features an exceptionally long, 13.2-inch triple-fan thermal solution that was almost too long for our graphics-card test bed.

XFX’s RX 7700 XT has a simple, almost all-black exterior. The card feels solid throughout, and the enclosure and PCB lack the noticeable flex that we see with more affordable thermal solutions, giving the XFX Speedster a premium feel.

Otherwise, the RX 7700 XT created by XFX has a fairly standard feature set. It has a nearly universal rear I/O panel configuration with one HDMI port and three DisplayPort jacks, as well as two PCI Express (PCIe) eight-pin power connections on the card. The card’s board power is listed as only 245 watts, but these two PCIe connections should provide at least 300 watts combined, providing plenty of power for overclocking.

4.Nvidia GeForce RTX 4070 Super

Last update on 2025-03-09 / Affiliate links/images from Amazon.

The Nvidia GeForce RTX 4070 Super is a graphics card that has a lot of expectations surrounding it since its announcement at CES 2024, and if you haven’t upgraded your graphics card in a minute and have been waiting for a sign, this release is what you’ve been looking for, whether you buy it or go with a competing card from AMD or Intel.

In terms of what you get for the same money, you will get significantly more SMs for more processing power, as well as a slightly higher base clock speed. Unfortunately, we’re still stuck with only 12GB GDDR6X VRAM, which limits this card’s 4K potential. If you want the best 4K graphics card, you might have to wait until January to see what the Nvidia GeForce RTX 4070 Ti Super and Nvidia GeForce RTX 4080 Super have to offer.

In terms of what you get for the same money, you will receive significantly more SMs for more processing power, as well as a slightly faster base clock speed. Unfortunately, we’re still stuck with only 12GB of GDDR6X VRAM, limiting this card’s 4K potential. If you want the best 4K graphics card, you may have to wait until January to find out what the Nvidia GeForce RTX 4070 Ti Super and Nvidia GeForce RTX 4080 Super have to offer.

Even when DLSS is considered, Nvidia falls behind the RX 7800 XT in terms of overall gaming performance. And, once AMD releases its own frame generation technology for FSR in the coming weeks, Nvidia’s advantage from DLSS 3 with Frame Generation will be even tighter. On this point, gamers will have to ask themselves more difficult questions than anyone else, and the price of the RX 7800 XT alone may be enough to tip the scales in their favor.

Still, it’s difficult to argue that Nvidia hasn’t delivered an absolutely fantastic card with the Nvidia GeForce RTX 4070 Super.Nvidia’s latest offensive in the graphics card market has begun with the introduction of new RTX 40-series Super GPUs. The GeForce RTX 4070 Super, the first of these new cards to hit store shelves, provides significant performance improvements over the “mundane” RTX 4070 at no extra cost. That’s right, it keeps the $599 price tag, giving gamers significantly higher frame rates (and value) than a card released less than a year ago. What is not to love? Not surprisingly, the Nvidia GeForce RTX 4070 Super is the clear Editors’ Choice.

The RTX 4070 Super, released less than a year after the Nvidia GeForce RTX 4070, features no major architectural changes. To boost the performance of this graphics card, Nvidia chose a more straightforward approach: increasing the number of cores. The original RTX 4070 was based on an Nvidia AD104 GPU die with 7,680 CUDA cores, but only 5,888 of them were active.

The GeForce RTX 4070 Super is also based on the AD104 GPU die, but has 7,168 active cores, representing a 22% increase. This is accompanied by a corresponding increase in other shading resources, with the new card employing approximately 22% more texture mapping units (TMUs), raster operation processors (ROPs), and ray-tracing (RT) cores. Exact figures are shown in the chart below.

5.AMD Radeon RX 7800 XT

Last update on 2025-03-09 / Affiliate links/images from Amazon.

AMD’s Radeon RX 7000-series graphics cards have provided fierce competition to Nvidia’s GeForce RTX 40-series GPUs. You’ve probably noticed a significant gap in AMD’s product lineup, with nothing between the $269 AMD Radeon RX 7600 and the $899 AMD Radeon RX 7900 XT. It is astonishing that this situation has persisted for so long. The AMD Radeon RX 7700 XT and Radeon RX 7800 XT, however, are both excellent new GPUs, so the wait was probably worthwhile.

When AMD introduced its Radeon RX 7000-series graphics cards, it identified three technologies as key components of the new architecture. The first was a new microarchitecture called “RDNA 3,” and the second was a new manufacturing process. Both of these were predictable, as such changes occur almost every generation. I discussed them in my launch review of the AMD Radeon RX 7900 XTX.

The third technology is a new chiplet design that debuted with the first AMD Radeon RX 7000-series graphics cards. However, chiplets do not power every Radeon RTX 7000-series card. Following the release of the chiplet-based Radeon RX 7900 XTX and the similar Radeon RX 7900 XT, AMD introduced the Radeon RX 7600, which features a monolithic chip design. This raised the question of what future cards in the series would look like. Now we have the answer.

The Radeon RX 7800 XT uses a chiplet design similar to the RX 7900 XTX and RX 7900 XT. It is built around one large chip known as the Graphics Chip Die (GCD). It measures 200mm2 and has 60 compute units, giving it 62.5% more resources than the GCD in the 7900 XTX.
In terms of cache, the RX 7800 XT includes four Memory Cache Dies (MCDs) measuring approximately 37mm2 each, for a total combined area of around 150mm2. Each of these MCDs has a 64-bit memory interface and 16MB of L2 cache, giving the RX 7800 XT an overall 256-bit memory interface and 64MB of L2 cache.

The chart above provides a more detailed breakdown of the number of ray accelerators, AI accelerators, TMUs, ROPs, and shaders. When the amount of resources decreases, we often see a slight increase in clock speed to compensate for the difference in shaders. This is not the case with the Radeon RX 7800 XT, which instead shows a slight drop in clock speed. It has a maximum boost clock of 2,430MHz, compared to the Radeon RX 7900 XTX’s 2,500MHz.

AMD provided us with a Radeon RX 7800 XT for review, which was based on an internally developed reference design. This card has two internal eight-pin PCIe power connections, three DisplayPort connectors, and an HDMI port on the rear I/O panel. The reference card uses a dual-fan thermal solution that’s slightly wider than a standard dual-slot card.

Best graphics cards performance results

The following charts contain data from tests that have been carried out over the last few months. We try our best to test everything at 1080p medium and ultra, but only the fastest cards are tested at 1440p and 4K. The first chart displays the geometric mean (i.e., equal weighting) for all 19 games for each resolution and setting. The third chart focuses on ray tracing performance in eight games, while the second chart displays performance in eleven rasterization games. For those who prefer to view all of the data, we also have the 19 individual game charts.

Only three of the nineteen games—Borderlands 3, Far Cry 6, and Total War: Warhammer 3—do not have a contemporary temporal upscaling solution (DLSS, FSR 2/3, or XeSS). Nvidia’s DLSS 2 has been available since mid-2019, AMD’s FSR has been available for more than two years (FSR 2.0 has now passed the year mark), and Intel’s XeSS was formally introduced in October 2023. Ten of the games in our test suite support FSR 2 or later, five support XeSS, six support DLSS 3, and sixteen support DLSS 2 or later.