Do Not Buy: NVIDIA RTX 5070 Ti GPU Absurdity (Benchmarks & Review)
jimmy_thang
February 21, 2025
We review the 5070 Ti’s value, gaming performance, ray tracing, efficiency, thermals, acoustics, and more
The Highlights
- The 5080 performs generally 9% to 16% better than the 5070 Ti
- The 5070 Ti is selling upwards of $1,000+ and performs similarly to a 4080
- We think you should not buy a 5070 Ti right now
- Original MSRP: $750
- Release Date: February 20, 2025
Table of Contents
Our
fully custom 3D Emblem Glasses celebrate our 15th Anniversary! We hand-assemble these on the East Coast in the US with a metal badge, strong adhesive, and high-quality pint glass. They pair excellently with our
3D 'Debug' Drink Coasters. Purchases keep us
ad-free and directly support our consumer-focused reviews!
Intro
You should not buy the NVIDIA RTX 5070 Ti (beware of scalped prices), or really, probably not any of the high-end cards right now. The market is in chaos and conditions are at an all time low for consumers. It benefits this entire industry, including us, if everyone mindlessly consumes, but we’re telling you not to be a part of that cycle. This is not the time to buy an NVIDIA video card. The company says the 5070 Ti (beware of scalped prices) will be $750, but we don’t believe it. In fact, some partner models have already been spotted at $1,000, with plenty of others coming in at $850 to $900. We heard of one that’ll have an MSRP above $1,000. That’s more than a 4080 (watch our review), which is insane.
NVIDIA calls it the RTX 5070 Ti, but it’s really more like an RTX 4080 (beware of scalped prices) V3 -- or V4 if you count the unlaunched 4070 Ti which was a 4080 12GB.
Editor's note: This was originally published on February 19, 2025 as a video. This content has been adapted to written format for this article and is unchanged from the original publication.
Credits
Test Lead, Host, Writing
Steve Burke
Testing
Patrick Lathan
Testing, Editing
Mike Gaglione
Testing
Jeremy Clayton
Camera, Video Editing
Vitalii Makhnovets
Writing, Web Editing
Jimmy Thang
The TLDW up front is that the 5080 (beware of scalped prices) is about 12-16% ahead of the 5070 Ti in some of the 4K gaming scenarios we tested, the 5070 Ti is about 28-35% ahead of the 4070 Ti at 4K, but down in the 20-28% range commonly at 1440p. The lead of the 5070 Ti over the 4070 Ti Super, which is the closest recent price neighbor, is only 7.8% to 20% in a lot of 4K situations, or commonly 12-16%. One of the worst scenarios we saw was an impressively bad 3.9% uplift from the 4070 Ti Super to the 5070 Ti at 1440p.
You should not buy this video card. You shouldn’t even read this article. We’ll save you the time: NVIDIA has become fat and monopolistic and AMD has decided it won’t compete in the high-end, and Intel is not there yet, leaving everyone with a worse ecosystem.
The 5080 (beware of scalped prices) to the 4070 is a difference that has now been subdivided several times. Like the 4080 (beware of scalped prices) and 4070 Ti (watch our review), which was a 4080 12GB originally, already subdivided it. Likewise, the 4080 Super subdivided it further. The 4070 Ti Super (read our review) subdivided that again, and now, the 5070 Ti subdivides it once more.
There are so many subdivisions in this block that it’ll have an HOA by next week...
In one of our tests, we saw an average gap between cards in this upper-end of 6.7 to 8 FPS card-to-card, depending which you want to count. That really means that the only relevant factor is price (except for maybe VRAM in the few places it diverges, where that’ll matter for professional users). But even VRAM has mostly been split into the 90-class and everything else.
The 5070 Ti is often the same performance as the 4080, which is the same as the 4080 Super. The improvement on the 4070 Ti Super, which was close to the same MSRP, is hardly meaningful. This aligns with NVIDIA’s launch strategy of either leveraging its monopolistic positioning or boring the ever-living f*** out of people with stagnated change. Meanwhile, it appears that AMD is sticking to its failed strategy of MSRP = NVIDIA - $50. Let’s get into the rest of the review.
RTX 5070 Ti Overview
We’re taking a look at the ASUS RTX 5070 Ti Prime model, which we assume is named like that because ASUS is trying to get Logan Paul’s attention. Or maybe it has electrolytes, which is what the gamers crave.
The RTX 5070 Ti has 8960 CUDA Cores, 16GB of GDDR7, and a 256-bit bus. The memory spec is identical to the RTX 5080 (read our review), with the 5080 having more CUDA cores but being a significant cut-down from the 5090 (read our review), which is more than what we’ve typically seen.
We have the full specs in a separate article here.
RTX 5070 Ti Pricing
Pricing is going to be the biggest problem for this GPU. Supply has been non-existent for the 5090 and 5080 and prices have skyrocketed into true FOMO targeting territory. We posted a video about the fake prices just a few days ago.
And in time for that, the 5070 Ti will launch with its own glut of pricing issues. We know one board partner has an actual MSRP north of $1,000 for one of its models. This gets worse considering the lower overall price of $750, which means that the proportional hike over expectation becomes untenable.
There’s really not a lot of ground to cover here that we didn’t in the intro, so let’s just get into the data.
RTX 5070 Ti Game Benchmarks
For the testing here, we had to remove a couple of cards from the charts to make space. Specifically, we removed the 3090 (watch our review) and 3090 Ti (watch our review). If you want the data for those cards, check our 5080 review as it has all that.
FFXIV 4K
In Final Fantasy 14 at 4K, the RTX 5070 Ti landed at 97 FPS AVG with lows at 83 and 80. That’s the same as an RTX 4080 which, under normal conditions, might sound like a generational improvement on the arbitrarily named GPUs. Instead what we have is total generational stagnation because the 5070 Ti will regularly sell at prices, at least for now, approaching what the 4080 was. Even at only a slight elevation to $850, the card is simply too close to the last generation performance equivalent.
The 7900 XTX outperforms the 5070 Ti by 7.4%, with the 5080 16% ahead and the 5090 at 87% ahead. As for the 7900 XT Hellhound, which was the cheapest partner model for a while, that’s at 82 FPS.
Generationally by name, the 5070 Ti improves on the 4070 Ti by 34%, but the 4070 Ti Super is the more recent price replacement. The lead over the 4070 Ti Super is just 13%. That’s stagnation. We have the RTX 2070 Super (watch our review) in a future test, but the 1070 (watch our review) is here if you want something older. The improvement is 332% over Pascal, now 9 years old.
FFXIV 1440p
At 1440p, the 5070 Ti ran at 187 FPS AVG. 1% and 0.1% lows are where they should be based on other results, so there’s nothing exceptional here. The 5070 Ti outranks the 4070 Ti Super by just 15 FPS, or only 9%. Sadly, this is one of the better ones, but that’s still brutal and is one of the most boring improvements possible when considering the MSRP similarity -- and that’s before you go get ripped off for more.
As for other 70-class cards: The 5070 Ti’s uplift over the 4070 Ti is 24%, then 72% on the 3070 Ti (watch our review), 94% on the 3070 non-Ti (watch our review), and 307% on the 1070. The uplift has diminished for each of these as resolution came down to 1440p, despite the 5070 Ti not being anywhere close to a CPU ceiling.
AMD’s RX 7900 XTX (watch our review) leads the 5070 Ti by 13%, up from 7% at 4K. The 5070 Ti also leads the 7900 XT by just 9% here.
FFXIV 1080p
At 1080p, things get even worse. The 5070 Ti leads the 4070 Ti Super by only 6%, or just 15 FPS average. Over the 4070 Ti non-Super (watch our review), which remains an absurd distinction, we’re only seeing a 16% advantage on the 5070 Ti. No wonder NVIDIA wants everyone to type the framerate into a calculator and multiply it artificiality by MFG.
The new card is just 10% over the 7900 XT, then 74% over the 3070 Ti.
Black Myth: Wukong - 4K
Black Myth: Wukong is up now, tested at 4K. We’re removing the experimental chart labeling and feel that this test has had enough public visibility to clear that bar.
The 5070 Ti ran at the exact same framerate as the RTX 4080. NVIDIA has basically re-released the 4080, and the price isn’t even that different -- and in some cases, not different at all. In other words, NVIDIA has now released three RTX 4080s. This is the third one. At some point, it just seems like this is too much.
The 7900 XTX matches the 5070 Ti here. The difference is irrelevant and unnoticeable.
Against the 4070 Ti Super’s 45 FPS AVG, the 5070 Ti leads by 13%, or 27% over the 4070 Ti non-Super and similarly over the 7900 XT. The lead over the 3070 Ti is 81%.
Black Myth: Wukong - 1440p
At 1440p, the “RTX 4080 v3” ran at 87.2 FPS AVG, with lows at 74.9 and 70.3. The third iteration of the 4080 isn’t the best one, though: The first version of the 4080 and second version of the 4080 are within error of each other and technically ahead of v3. But on the v3, aka the 5070 Ti, NVIDIA has enabled Multi-Fiat Generation to make it more interesting, though.
The RX 7900 XTX is about the same as the 5070 Ti here, with the 4070 Ti Super just below it. The 5070 Ti is only 10% ahead of the 4070 Ti Super and 20% ahead of the 4070 Ti, with a similar lead over the 7900 XT.
Black Myth: Wukong - 1080p
At 1080p, the 5070 Ti is again roughly tied with the 7900 XTX, although credit to NVIDIA for having improved 0.1% lows -- just not by an amount anyone would notice in play. The frametime pacing is still good on both.
The 4080 Super (read our review) and 4080 are again within error of each other and, although they’re outside of error vs. the 5070 Ti, the real error is the GPUs we launched along the way.
The 5070 Ti ends up leading the 105 FPS 4070 Ti Super result by a staggering, mind-blowing 9.7%, made only more impressive by the fact that this was already functionally the same as the 4070 Ti’s 100 FPS.
The 7900 XT (read our revisit) sits just below this.
Starfield - 4K
Starfield is up now. At 4K, The RTX 5070 Ti ran at 68 FPS AVG, this time allowing the RTX 4080 the important distinction of... being technically better.
The 5080 only leads the 5070 Ti by 12% here as well.
Troublingly for the 5070 Ti in this test, the 7900 XT nearly matches the RTX 5070 Ti, with the 7900 XTX ahead of it by 13%.
The 4070 Ti Super and 4070 Ti aren’t that different from each other in this one, either.
Starfield - 1440p
At 1440p, the 4080 Super and 4080 are exactly tied, with the 4080 Sub-Super, or as NVIDIA calls it, the “5070 Ti,” at 101 FPS AVG.
The 5070 Ti ends up basically tied with the 7900 XT. The 7900 XTX holds an advantage at 11% ahead. The 5070 Ti’s improvement, if you can call it that, over the 4070 Ti Super is an impressively boring 3.9%, with the lead over the 4070 Ti at 11%. That’s impressive -- mostly because we’re impressed NVIDIA could make something so impressively stagnant.
Starfield - 1080p
At 1080p, Starfield has the 5070 Ti below the XTX, which has a slight lead at 6%. More impressively, the 4070 Ti Super ran at 120 FPS AVG. Between the 5080, the 4080s, the 5070 Ti, and the 4070 Ti Super and 4070 Ti, NVIDIA has managed to make a video card for seemingly every individual framerate between 110 FPS and 132 FPS. Why they’d do this, we have absolutely no idea.
Dragon’s Dogma 2 - 4K
Dragon’s Dogma 2 is up now. This one has the ASUS RTX 5070 Ti Electrolyte, which is what the games crave, at 73.6 FPS AVG. That’s right between the 4080 Super and 7900 XTX, making the 5070 Ti the first NVIDIA RTX 4080 Super-Super on the market. We won’t be impressed until they launch the triple-S-tier.
The 5080’s lead over the 5070 Ti is just 15%. That small of an uplift is going to weigh on NVIDIA.
The 5070 Ti leads the 4070 Ti Super by 18% and the 4070 Ti by 35%. The 7900 XT sits between the two 4070 Ti variants.
Dragon’s Dogma 2 - 1440p
1440p reduces the 5070 Ti’s relative ranking, pushing it below the 4080 and 4080 Super, the two of which are tied and within error of each other. The 4070 Ti Super’s 106 FPS AVG result also makes for a boring positioning of the 5070 Ti. Again, we’re back to a card for every couple FPS. More than ever, this means price matters more.
The 7900 XTX leads the 5070 Ti by about 6 FPS here.
Dragon’s Dogma 2 - 1080p
At 1080p, the RTX 5070 Ti’s 151 FPS AVG basically ties it with the 7900 XTX, which itself is tied with the 4080 (which is tied with the 4080 Super, which is a waste of chart space and so isn’t shown).
Generationally, the 5070 Ti leads the 4070 Ti Super by 10% and 4070 Ti by 18%.
Resident Evil 4 - 4K
Resident Evil 4 is up now, first rasterized. The 4K test has the 5070 Ti at 107 FPS AVG, with the 5080 at 122 and leading by 15%. Both are behind the RX 7900 XTX at 126 FPS AVG.
The 5070 Ti is ahead of the 4070 Ti Super by 20% and 4070 Ti by 34%. The 7800 XT (watch our review) is between these and the regular 4070, followed by the older 3070 Ti at 53 FPS AVG.
Resident Evil 4 - 1440p
At 1440p, the 5070 Ti is just below the 4080 FE. The 5080 leads the 5070 Ti by 13%.
NVIDIA basically took what previously would have been a 51 FPS gap from 224 to 173 FPS between the 4070 Ti Super and 5080, then split the difference with the 5070 Ti.
The 7900 XTX is ahead by 17%, with the XT just below the 5070 Ti.
NVIDIA RTX 5070 Ti Ray Tracing Benchmarks
Visit our
Patreon page to contribute a few dollars toward this website's operation (or consider a
direct donation or buying something from our
GN Store!) Additionally, when you purchase through links to retailers on our site, we may earn a small affiliate commission.
We’re moving to ray tracing testing now. We’ll start with Black Myth, which is heavily NVIDIA favored. Then we’ll look at some that are mixed or lighter workloads.
Ray Tracing: Black Myth Wukong 4K
At 4K with upscaling as defined in the chart header, the 5070 Ti ran at 52 FPS AVG and tied the RTX 4080 and 4080 Super. The 5080’s 59 FPS result had it 13% ahead. Over the 4070 Ti Super, we see a 15% lead for the 5070 Ti. AMD gets absolutely crushed in this test.
Ray Tracing: Black Myth Wukong 1080p
At 1080p and still ray traced, the RTX 5070 Ti ran at 112 FPS AVG and led the 4080. The 5080 is improved on the 5070 Ti by 9%. The proximity of cards from the 4070 through the 5080 is crazy, though: We have the 4070 at 80 FPS, then the Ti at 92, then the Ti Super at 95, then the 4080 at 106, then the 4080 Super which isn’t shown but the same, then 5070 Ti at 112, then the 5080 at 122 FPS.
Ray Tracing: Dragon’s Dogma 2 4K
Dragon’s Dogma 2 at 4K with ray tracing is next.
In this one, the 7900 XTX is more competitive and lands at 66 FPS AVG, which is between the 5080 and 5070 Ti. It’s unfortunate that AMD gets crushed so hard in some of the other games, like Black Myth: Wukong because it does OK in the ones that are less crazy intensive. The 5070 Ti ends up basically tied with the 4080 Super, which is basically tied with the 4080. The jump over the 4070 Ti Super’s 54 FPS AVG is 18% here.
Ray Tracing: Dragon’s Dogma 2 1440p
1440p positions the 5070 Ti between the 4070 Ti Super and 4080. It’s not clear why this card needs to exist, but it does. The improvement against the 4070 Ti Super is just 10.8%. The 7900 XTX runs at 108 FPS AVG and is between the 4080s and the 5080.
Ray Tracing: Dragon’s Dogma 2 1080p
At 1080p, the 5070 Ti ran at 131 FPS AVG and was roughly tied with the 7900 XTX. The 4080 leads the 5070 Ti here by a few percent. The 5070 Ti leads the 4070 Ti Super by 9%, followed by the 7900 XT in the middle, then the 4070 Ti at 16%.
Ray Tracing: Resident Evil 4 4K
In Resident Evil 4 at 4K with ray tracing and upscaling, the RTX 5070 Ti ran at 118 FPS AVG and tied the 4080 and 4080 Super exactly. It’s within run-to-run variance and error. The 7900 XTX leads in this lightweight RT workload with a 14% advantage, posting a big difference from the heavier Cyberpunk and Black Myth workloads that we run.
Ray Tracing: Resident Evil 4 1440p
At 1440p with the same settings, we see a similar lineup. The 5070 Ti is again within error of the 4080 and 4080 Super, which are the same. The 7900 XTX is slightly improved. The 4070 Ti Super is slightly behind. This is uninteresting.
RTX 5070 Ti Efficiency Benchmarks
Efficiency remains a relatively new test for us to include in each review, so we haven’t re-run the 4070 cards for efficiency yet.
Efficiency: FFXIV 4K
In Final Fantasy 14 at 4K, the 5070 Ti landed at 0.37 FPS/W. It pulled 264W in this test, approaching its TDP spec. The 5080 ends up slightly more efficient, with the 4080 Super equivalent. The 7900 XTX is at a large disadvantage here due to its power consumption.
Efficiency: FFXIV 1440p
In Final Fantasy 14 at 1440p, the 5070 Ti ended up at 0.73 FPS/W, which has it just below the 4080 and 5080. The 5070 Ti improves on the 4060-class cards, but also is significantly more efficient than the 7900 XTX.
Efficiency: Black Myth: Wukong 1080p
In Black Myth Wukong at 1080p, we found the 5070 Ti to be between the 4080 Super and 5080 for efficiency. The 7900 XTX shows again that this is its weakness, but its particular performance deficiency in Black Myth in general is hurting it disproportionately.
Efficiency: F1 24 4K
In ray tracing performance with F1 24, we found the 5070 Ti to perform about the same as the 4080. The 7900 XTX scores significantly lower. Overall, here you’re seeing much lower numbers than in some of the other charts and that’s because this is a heavy ray tracing workload tested at a higher resolution.
RTX 5070 Ti DLSS4/MFG
All of this means that NVIDIA is basically just selling you an RTX 40 series card, maintaining elevated prices, and doing so while pushing DLSS4 and MFG as the only real differentiating factor on the 50-series cards.
We already have a piece up talking about the new Transformer model versus the prior CNN model for DLSS, which you can check out for a deep dive. We also have a video with an image quality comparison, including a frame-by-fake-frame break-down of MFG, where we analyze the AI or synthetic frames against the keyframes, or the native frames. We do this at 2X and with MFG 4X.
The improvements in the Transformer model over the older CNN model for image quality are apparent; meanwhile, the generated frames serve their purpose of smoothing, but sometimes look bad. What they don’t do is turn a low frame rate like 20 FPS into something that is instantly playable. But as far as image quality, it’s highly situational; in some situations, DLSS ends up better than native because game developers have decided to ruin their games with terrible default options, which shouldn’t happen, but there are also a lot of scenarios where it looks awful. We also found bugs in the driver override features that NVIDIA has pushed to the public. This includes sometimes it incorrectly running the generated frames in the wrong places.
RTX 5070 Ti Thermals
This quick thermal chart at steady state during a looping 4K Port Royal workload and it shows the GPU and memory results. We don’t have other 5070 Ti cards to compare, so we can’t produce a like-for-like comparison. The 5080s are only here for reference.
The 5070 Ti came in 2 dBA quieter than the Zotac 5080 Solid (read our review) and was warmer while operating at a significantly lower reported board power during the test. The ASUS cooler is favoring noise here, but is also just not particularly effective. Overall, the performance is fine -- this is more than acceptable and well below throttle territory -- but this does seem to be one of the lower cost coolers for its size.
RTX 5070 Ti Acoustics
Acoustic testing is up next. We’ll keep this brief. As a quick positive, our chamber is about to get way better for future testing -- or more accurately, our microphone equipment. We’ve recently learned that with a microphone upgrade, we can bring down the noise floor closer to 5dBA, which is crazy exciting. Currently though, we’re on our more economical equipment.
With the current noise floor of 14-14.5 dBA, we measured the ASUS 5070 Ti Prime at 29.2 dBA with this frequency spectrum plot. We observed one spike at around 300 Hz. Otherwise, it follows what we have seen in other cards. The falloff begins at 2000 Hz, levels briefly at 3500 to 5000, then continues the path. The loudest range is 1000 Hz to 1800 Hz, aside from the 300 Hz spike.
These two added lines show the Zotac 5080 and NVIDIA 5090 FE (beware of scalped prices). The coolers should track about the same as long as the fan RPM is the same between the other 5080 and 5090 alternatives by the same vendor. The NVIDIA card had a spike around 180 Hz and another at around 380. The Zotac card had one in the 430 Hz range. Broadly, the NVIDIA FE is louder than both of these.
RTX 5070 Ti Conclusion
Grab a
GN15 Large Anti-Static Modmat to celebrate our 15th Anniversary and for a high-quality PC building work surface. The Modmat features useful PC building diagrams and is anti-static conductive. Purchases directly fund our work! (or consider a
direct donation or a
Patreon contribution!)
This will change in the future as prices move, hopefully, but for now, we don’t recommend buying this card. Honestly, it’s probably just a good idea to wait in general right now. But let’s just recap why that is as quickly as possible:
Broadly speaking, we found that the 5080 performs in the range of 9% to 16% better than the 5070 Ti, depending on the resolution and game. The 7900 XTX is often better than the 5070 Ti in rasterized testing, with the range in our games spanning equivalence to 17% at the high end, but more commonly 6% to 13%. There were some instances of regression for the 7900 XTX vs the 5070 Ti, but as has been the case, it only got really crushed in Black Myth or Cyberpunk-type ray tracing. The 5070 Ti only improves on the 4070 Ti Super by a range of 2.2% to 20%, often 12-16% if you’re looking at the heavier resolutions. That is not a good improvement for a generational jump and the actual street price we expect as compared to the 4070 Ti Super’s original street pricing, especially with the instances below 10% performance. It’s an awful value. Remember that the Super series was regularly in stock for its actual MSRP.
As we’re reviewing this, we can’t see the actual street price of the 5070 Ti in advance and we don’t review the future as we don’t know where the prices will land at launch, but based on what we’ve seen from the 5080 and 5090, our strongest recommendation right now is to just generally wait to buy a video card and let the market calm down.
Finally, the 5070 Ti is equal to the 4080 Super. We don’t care what the 4080 MSRP was, because the 4080 Super effectively overwrote it. That was $1,000 and was actually available at around $950 to $1,050 for much of its recent life.
For AMD’s part, it really needs to not screw its new GPUs up. Typically, whatever AMD says its price is, it comes down in about 1 quarter, because they usually push it way too high. It’s really disappointing that AMD said it’s not going to be targeting high-end GPUs this generation.
Overall, we think it’s not a good time to buy a GPU and wouldn’t recommend buying a 5070 Ti.