If you play games with upscaling enabled and want to upgrade to an upper mid-range graphics card, should you purchase the GeForce RTX 5070 Ti or the Radeon RX 9070 XT?
That’s what we’ll be diving into today, comparing GeForce vs Radeon using DLSS 4 and FSR 4 upscaling, respectively. We are also pulling data from our previous benchmark reviews of the RTX 5070 and [RX 9070](https://www.techspot.com/review/3044-nvidia-rtx-5070-vs-amd-radeon-9…
If you play games with upscaling enabled and want to upgrade to an upper mid-range graphics card, should you purchase the GeForce RTX 5070 Ti or the Radeon RX 9070 XT?
That’s what we’ll be diving into today, comparing GeForce vs Radeon using DLSS 4 and FSR 4 upscaling, respectively. We are also pulling data from our previous benchmark reviews of the RTX 5070 and RX 9070 to build a clear picture of how the mid-range market is stacking up.
At the moment, there’s a price gap between the two cards, though not as wide as you might expect. The RTX 5070 Ti is holding steady at its $750 MSRP, with several base models widely available at retailers like Newegg. The RX 9070 XT, meanwhile, is hovering around $650 – about $50 above its intended MSRP. The question is: does that price difference meaningfully change the outcome?
Our goal today is to provide a different data set from our day-one reviews, which primarily focus on native performance at various resolutions. Native performance data is included today as well, but the emphasis is on key upscaling modes at 1440p: Quality, Balanced, and Performance.
We want to see what kind of frame rates you can expect in a realistic gaming setup, whether DLSS 4 or FSR 4 scales better, and which settings are required to match performance output between the two models.
As with any upscaling benchmarks, it’s important to understand the visual differences between DLSS 4 and FSR 4, because they do not produce identical output. This means performance testing is not a perfect one-to-one comparison. We’ve published a lot of in-depth content on TechSpot that explains how these upscalers differ, and those articles serve as valuable companions to this kind of performance testing.
Test System Specs
For testing today, we’re using the MSI Ventus 3X OC RTX 5070 Ti and the Sapphire Pulse RX 9070 XT, along with the MSI Vanguard SOC RTX 5070 and the Asrock Steel Legend OC RX 9070. Since these are different models, some with factory overclocks, all GPUs have been set to their base clock specifications, which is the standard way we benchmark.
The CPU powering our test machines is the Ryzen 7 9800X3D, paired with 32 GB of DDR5-6000 CL30 memory. The latest game updates, Windows updates, and drivers were applied as of writing (version 581.57 for Nvidia and 25.10.1 for AMD). In both the GeForce and Radeon control panels, we enabled the global override features that upgrade games to the latest versions of DLSS 4 and FSR 4 respectively.
Let’s get into it.
Gaming Benchmarks
Horizon Zero Dawn Remastered
Horizon Zero Dawn Remastered runs slightly better on the Radeon model, with the 9070 XT delivering 4% higher performance using native rendering and extending that lead to 6% with Quality upscaling. Both GPUs provide a smooth high-refresh-rate experience when upscaling is enabled, and there’s little benefit in using settings below Quality. In this title, the 5070 Ti performs similarly to the RX 9070.
Ratchet & Clank: Rift Apart
Testing was conducted using the Very High preset with ray tracing enabled in Ratchet & Clank, which these GPUs handle with ease. The RTX 5070 Ti is the faster card here, offering 16% higher performance than the 9070 XT in native rendering and 15% more using Quality upscaling.
To achieve equivalent frame rates, the GeForce card can use DLSS Quality mode, while the Radeon card must drop to FSR Performance mode, which results in noticeably lower image quality.
The Last of Us Part I
Using native rendering, the 5070 Ti and 9070 XT perform almost identically in The Last of Us Part I. However, with Quality upscaling enabled, the GeForce model takes a small 3% lead. At most, a 5% advantage was observed for the 5070 Ti using Performance mode. Overall, both GPUs deliver a similar experience, sitting around 10% ahead of the RX 9070.
Spider-Man 2
Spider-Man 2 favors the RTX 5070 Ti, especially with upscaling enabled. The GeForce card is 11% faster than the 9070 XT using native rendering and 19% faster with Quality upscaling. There’s also a substantial 30% lead in 1% lows. With ray tracing enabled, DLSS 4 scales more effectively than FSR 4 in this game, making the experience notably better on the 5070 Ti.
Assassin’s Creed Shadows
Assassin’s Creed Shadows is a clear win for the RX 9070 XT. It delivers 15% more performance using native rendering, narrowing to a 10% lead with Quality upscaling, but still clearly ahead. This gap only shrinks when using Performance mode upscaling – a setting that most players are unlikely to choose at 1440p. In general, upscaling is less effective in this title compared to others in terms of performance gains over native rendering.
The Last of Us Part II
While performance was nearly identical between the 5070 Ti and 9070 XT in the first game, the The Last of Us sequel tells a different story. The Radeon model leads by 13% using native rendering and 9% with Quality upscaling. The RX 9070 is also faster when upscaling overall, with a 5% performance advantage.
Ghost of Tsushima
Ghost of Tsushima performs better on the RX 9070 XT, coming in 8% ahead of the 5070 Ti using Quality upscaling, with similar results when using native rendering. This is another title where the 5070 Ti is closer in performance to the RX 9070, sitting just 3% faster than that model.
Stalker 2
Stalker 2 is one of the more demanding games in this test suite and is a good example of FSR 4 scaling slightly better on Radeon hardware than DLSS 4 on GeForce. Using native rendering, the 5070 Ti is 12% faster than the 9070 XT, but this advantage drops to just 3% when Quality upscaling is enabled. 1% lows also struggle to rise much above 60 FPS in the test area, regardless of upscaling mode, largely due to traversal stutter in Unreal Engine 5.
Marvel Rivals
Marvel Rivals is a close battle, with the 5070 Ti slightly ahead of the 9070 XT: there’s a 4% margin using native rendering and a 5% margin using quality upscaling. At lower upscaling settings like Performance, the two cards are basically offering the same experience
Kingdom Come: Deliverance II
Kingdom Come is a win for the RTX 5070 Ti, which delivers 10% more performance than the 9070 XT using native rendering and 12% more with Quality upscaling. However, this lead mainly shows in average frame rates – 1% lows are much closer, meaning the Radeon card delivers more consistent frame pacing when upscaling is used. Natively, the 5070 Ti also holds a lead in 1% lows.
Star Wars Outlaws
Star Wars Outlaws continues to show how FSR 4 can scale more effectively than DLSS 4. The 5070 Ti is 8% faster than the 9070 XT using native rendering, but becomes 8% slower when using Quality upscaling. This is because the RX 9070 XT gains a 39% boost from native to Quality upscaling, while the 5070 Ti only gains 18%. With upscaling enabled, the 5070 Ti performs closer to an RX 9070, which wouldn’t be clear from native-only data.
Hunt: Showdown
Hunt: Showdown is essentially a tie, with the RX 9070 XT leading by just 3% whether upscaling is enabled or not. At launch, the RTX 5070 Ti was significantly faster, but recent driver updates have improved performance on the 9070 XT, bringing it back in line in this still-popular title.
God of War Ragnarök
In God of War Ragnarök, both GPUs perform similarly using native rendering, but the RX 9070 XT pulls ahead by 5% when using Quality upscaling. The 5070 Ti remains faster than the RX 9070 at equivalent settings and is up to 13% faster using native rendering, though it typically lands between the RX 9070 and 9070 XT.
Stellar Blade
Stellar Blade is still a bad game for Radeon, with the RTX 5070 Ti delivering its best result yet relative to the 9070 XT. Using native rendering it’s 29% faster, and a similar 27% margin is seen using Quality upscaling. The GeForce card also has much higher 1% low performance in the area we test, a 30% difference using quality upscaling, so AMD has some work to do to improve performance in this title.
F1 25
F1 25 is another example of the fastest card changing depending on the metric that we look at. Natively, the 5070 Ti is 6% faster than the 9070 XT, but this switches to the 9070 XT being 2% faster when viewing average frame rates using Quality upscaling. Additionally though, the 5070 Ti is 9% faster in terms of 1% lows, so the GeForce model is more consistent and this is what ultimately makes it the better choice for this game.
Cyberpunk 2077
Cyberpunk 2077 using the ultra ray tracing preset is the heaviest example of ray tracing out of the games we’re testing today. The 5070 Ti is unsurprisingly 18% faster using this configuration when rendering natively, and 13% faster on average using quality upscaling.
But like we saw in F1 25, there is a larger difference in 1% lows, with the 5070 Ti having a 24% lead there using quality upscaling, so not only is the 5070 Ti faster it’s also the more consistent GPU. Once you get down to models like the RX 9070, frame pacing is noticeably worse as the card is only hovering around the 60 FPS mark so any dips below that have more impact.
Hogwarts Legacy
In Hogwarts Legacy using the Ultra preset, both GPUs deliver nearly identical performance. The 5070 Ti is slightly ahead natively, but once upscaling is used, results are effectively the same. This title is a clear tie in both average frame rates and 1% lows.
Borderlands 4
Borderlands 4 is another title that offers similar performance on both the RTX 5070 Ti and RX 9070 XT using the Badass preset. At most the 9070 XT was 2% faster using quality upscaling, but this margin is close enough we call it a tie.
Mafia: The Old Country
Mafia: The Old Country runs better on the RX 9070 XT, with a 9% performance lead using native rendering and 13% using Quality upscaling. In this title, the 5070 Ti performs closer to the RX 9070 rather than the 9070 XT, making this a solid result for Radeon.
Dying Light: The Beast
Dying Light: The Beast strongly favors GeForce GPUs. The RTX 5070 Ti delivers 19% more performance than the 9070 XT using native rendering and 14% more with Quality upscaling. The 9070 XT needs at least Balanced upscaling to approach the 5070 Ti’s performance using Quality mode.
The Alters
Performance in The Alters is quite similar between both GPUs. The RTX 5070 Ti is 3% faster natively, while the RX 9070 XT is 4% faster using Quality upscaling. The Radeon model tends to produce better 1% lows at similar average frame rates, and FSR 4 appears to scale slightly better than DLSS 4 in this title.
Battlefield 6
The last title we looked at is Battlefield 6 in a custom bot match. Using native rendering at 1440p with the Overkill preset, the 5070 Ti was 5% faster than the 9070 XT. This lead extended to 9% when using quality upscaling and 11% using balanced upscaling, so DLSS is scaling a little better than FSR here. 1% low performance is closer between these models but typically the GeForce card offers the faster experience.
22 Game Average
Now it’s time to look at the 22 game average, calculated using the geometric mean. Starting with native rendering, the RTX 5070 Ti is 5% faster than the RX 9070 XT. This is similar to the margin we reported in our 55-game benchmark earlier this year, though slightly different compared to the more recent 9070 XT re-test, where we found the 9070 XT to be faster than it was at launch. In that review, across a 16-game average focused mostly on rasterized titles, the 9070 XT was about 3% ahead of the 5070 Ti. Today’s test sample combines both rasterized and ray-traced games.
However, when using Quality upscaling, the gap between the two GPUs shrinks – the 5070 Ti is only 2% faster, making it essentially a tie. Similar margins are seen in lower upscaling modes: 3% with Balanced and 2% with Performance, along with nearly identical 1% low results. Across a solid sample of games at 1440p, using typical upscaling settings, the RTX 5070 Ti and RX 9070 XT deliver very similar performance.
The data also shows that the 5070 Ti is 15% faster than the RX 9070 using native rendering and 9% faster using Quality upscaling. Typically, this means that the RX 9070 using Balanced upscaling will match the RTX 5070 Ti using Quality upscaling. The 5070 Ti is faster and offers better visual output, but the gap isn’t huge at 1440p.
Compared to the RTX 5070, the 5070 Ti is 27% faster using native rendering and 24% faster with Quality upscaling. There’s no DLSS 4 setting on the 5070 that can match the 5070 Ti using DLSS 4. In fact, what the 5070 achieves with Quality upscaling is similar to what the 5070 Ti delivers natively.
RX 9070 XT vs RTX 5070 Ti: Native Difference
When we look at the spread of results between the 5070 Ti and 9070 XT using native rendering, we do see more examples where the 5070 Ti is more than 5% faster. This was the case in 9 games, whereas the 9070 XT was more than 5% faster in just 4 games.
The 5070 Ti tends to perform better in ray-traced games like Cyberpunk 2077, Ratchet & Clank, and Spider-Man 2 – though that’s not universal, as games like Stellar Blade and Dying Light: The Beast don’t use ray tracing.
RX 9070 XT vs RTX 5070 Ti: Quality Upscaling Difference
Switching to Quality upscaling, results even out. There are 7 games where the 5070 Ti is more than 5% faster, 10 games where performance is effectively tied, and 5 games where the 5070 Ti is more than 5% slower. On average and across the spread of results, these two GPUs are evenly matched.
Which GPU Should You Buy?
With all that said and done, should you buy the RTX 5070 Ti or the RX 9070 XT? If both GPUs are sitting at their intended MSRP: $750 for the 5070 Ti and $600 for the 9070 XT, the Radeon graphics card is the obvious choice. With upscaling enabled, performance is practically on par with the 5070 Ti, but it costs 20% less. At those prices, the Radeon card delivers more frames per dollar and makes the most sense.
But, there are nuances to this discussion... the RTX 5070 Ti remains the more polished product overall. Nvidia’s DLSS 4 is supported in more games, offers better image quality in most titles, and older games can still leverage DLSS 2 or 3 – something AMD can’t always match. In games where FSR 4 is supported, the Radeon 9070 XT delivers a similar experience to the 5070 Ti. In games without it, the GeForce has a large image quality advantage. The 9070 XT doesn’t have the raw horsepower to compensate for weaker upscaling support.
Whether that matters depends entirely on what you play. If you stick to new releases, AMD has done a solid job delivering FSR 4 on launch day for big titles like Battlefield 6, and there’s no reason to believe that momentum is slowing. For players living in the present – not the past – the Radeon experience is largely the same. But if your library includes older favorites or niche titles, Nvidia’s broader DLSS ecosystem starts to look a lot more appealing.
Even so, pricing shifts the equation. At MSRP, the cost-per-frame advantage of the 9070 XT makes it a compelling option. Take Cyberpunk 2077 with Ultra ray tracing: the 5070 Ti is 13% faster on average with 24% stronger 1% lows – but it’s also 25% more expensive. That kind of ray tracing lead only really matters if you’re chasing extreme use cases like path tracing, where the 5070 Ti is more usable. But for most gamers (even single-player fans) paying 25% more for better performance in a handful of niche scenarios hardly makes sense.
But current prices are closer. Right now, the 9070 XT is around $650, while the 5070 Ti is $750. That narrows the gap to just $100, about 13% cheaper instead of the 20% it should be. At that price, the Radeon 9070 XT is less of a slam dunk but still reasonably priced considering its similar 1440p upscaling performance. It is decent value, just not the great value it would be at $600.
Ironically, the best deal in all of this isn’t the 5070 Ti or the 9070 XT – it’s the Radeon RX 9070. Sitting at its $550 MSRP, it makes the 5070 Ti 36% more expensive for only 9% more performance when using upscaling at 1440p. On a cost-per-frame basis, the RX 9070 is 20% cheaper, which is exactly the pricing spread we should see between the 9070 XT and 5070 Ti at MSRP.
Sure, there are edge cases where the 5070 Ti crushes the RX 9070. But in most titles, the Radeon stays in the same performance ballpark, and with small tweaks, it delivers similar results, all while costing $200 less and offering the same VRAM capacity.
And then there’s Australia (and likely other places), where the value story tilts even harder toward AMD. The RTX 5070 Ti starts at roughly $1,400 AUD, the RX 9070 XT at $1,100 AUD... the 20% gap Nvidia should be facing everywhere. But the RX 9070? Just $900. That makes the 5070 Ti a staggering 55% more expensive for only a 9% performance uplift. The 9070 XT is still a good buy versus the 5070 Ti, but the RX 9070 is the real mid-range hero in this lineup.