What makes for a great graphics card? I’ve spent a lot of time over the last few months thinking about that question, especially in the wake of Nvidia GPUs like the RTX 5090 and RTX 5080 which are good cards on their own, but don’t do much to justify their existence over what was already out there beyond the addition of multi-frame generation. The GeForce RTX 5060 Ti does buck the trend a bit, featuring the largest generational improvement out of the entire Nvidia Blackwell lineup.

This makes the RTX 5060 Ti the easiest current-generation Nvidia card to recommend, but that recommendation comes with a huge asterisk. You see, there are two versions of this graphics card out there, one with 8GB of VRAM and another, reviewed here, with 16GB. In a world where the Intel Arc B580 exists, it’s mind-boggling that Nvidia would launch an 8GB version of this card that’s still more expensive than Intel’s flagship. If you’re shopping for a 1080p graphics card, please don’t fall into the trap of buying the 8GB version.

Specs and Features

The Nvidia GeForce RTX 5060 Ti is built on the same Blackwell graphics architecture as the rest of the RTX 5000 lineup, this time with 4,608 CUDA cores across 36 Streaming Multiprocessors or SMs. That also means this graphics card has 152 Tensor Cores and 36 RT cores, which are used for DLSS and ray tracing, respectively. This configuration means its a bigger GPU than its predecessor, the RTX 4060 Ti, which had 4,352 CUDA cores across 34 SMs.

From there, the RTX 5060 Ti splinters into two different versions of the card, an 8GB and a 16GB model, both on a 128-bit bus. In fact, the actual GPU is identical between the two, with the only difference being the amount of VRAM available. For some people, this difference will be negligible, but if you’re looking to play intensive AAA PC games like Doom: The Dark Ages, 8GB could be a huge limiting factor, even at 1080p.

In a perfect world, there will be a $50 difference between the 8GB and 16GB versions of the RTX 5060 Ti, with them launching at $379 and $429, respectively. However, right now even 8GB versions of the card are starting around $400, with the 16GB card sitting anywhere from $500-$550 on Newegg. At that price, getting a graphics card with 8GB of VRAM simply doesn’t make sense, especially in a world where the Intel Arc B580 exists. Hopefully these higher prices normalize over the next few months and years.

It doesn’t help that there is no Founders Edition for the RTX 5060 Ti to help anchor prices. Instead, Nvidia is only selling the RTX 5060 Ti via third party manufacturers, just like the RTX 5070 Ti back in November. I was sent the PNY GeForce RTX 5060 Ti OC for review, which is a bog-standard dual-fan card with a modest 120MHz overclock. That’s less than a 5% boost to its clock speed, and shouldn’t make a huge impact on performance one way or the other. It also keeps the RTX 5060 Ti locked to the default 180W power budget, and it peaked at 181W in my testing. What’s interesting is that PNY doesn’t require the controversial 12V(2x6) connector as other Nvidia graphics cards. Instead, just a single run-of-the-mill 8-pin PCIe power connector is enough to power this RTX 5060 Ti. Though, there will likely be many versions of this graphics card that do require an adapter, but it should be included in the box.

DLSS 4

Just like the rest of the RTX 5000 lineup, the big selling point this time around is DLSS 4, and in particular Multi-Frame Generation, or MFG. Just like the Frame Generation tech that debuted with the RTX 4080 in 2022, MFG uses AI to create new frames between sets of rendered frames. The big difference now is scale, with the RTX 5060 Ti able to generate up to three interpolated frames for each natively rendered frame, greatly improving frame rate – though that comes with a catch.

Inserting interpolated frames into the display output naturally introduces a bit of latency, as the GPU is holding the rendered frame while it creates the AI frames. Luckily, Nvidia mitigates that through Reflex and the new AI Management Processor or AMP. The latter is a special bit of silicon on the GPU that takes over the responsibility of frame pacing from your CPU. And because it’s physically located on the same chip as, well, the rest of the GPU, it’s able to do this much more efficiently, shaving off some of the added latency and making the end result smoother.

Reflex, on the other hand, has been around for a while, and essentially zeroes out the render queue, and forces the CPU to only send data to the GPU when it’s ready to render a new frame, rather than stacking up a bunch of work and then sitting idle.

Even with these two methods, latency does still increase when MFG is enabled, but it’s not noticeable, especially if you’re already getting a decent frame rate when you turn it on. That’s the trick: You shouldn’t look at any frame generation trick as a magic bullet to improve your terrible frame rates, it’s just a tool to take an already good framerate and crank out extra frames to saturate high-end gaming monitors.

In Cyberpunk 2077, at 1080p on the Ray Tracing Overdrive preset, the Nvidia GeForce RTX 5060 Ti gets 83 fps and 29ms of latency with DLSS set to performance. Turn on 2x Frame Generation, and the frame rate jumps to 119 fps, with 36ms of latency. Then, with 4x MFG – so 3 AI frames for every “real” frame – the framerate jumps to 185 fps, with 42ms of latency. It’s a significant increase to latency, but it’s still low enough to not be super noticeable, thanks to the high starting frame rate.

Likewise in Star Wars Outlaws, the RTX 5060 Ti starts with 91 fps and 25ms of latency at 1080p, on the Ultra Preset and DLSS set to performance. With 2x frame generation, it jumps to 156 fps and 30ms of latency. Then, at 4x MFG, the numbers go up to 251 fps with 34ms of latency. The higher your frame rate is before frame generation is enabled, the better your experience is going to be.

Performance

At 1080p, the Nvidia GeForce RTX 5060 Ti has no problem handling most games with all the shiny features turned on. Across the suite of games I test GPUs with, I found that the RTX 5060 Ti was on average 23% faster than the RTX 4060 Ti, and 41% faster than the RTX 3060 Ti. That’s a far cry from the measly 11% improvement enjoyed by the RTX 5080 over the RTX 4080 Super.

I tested each graphics card on the most recent public driver, which means all Nvidia cards were tested on Game Ready Driver 576.02, and AMD cards were tested on AMD Adrenalin 25.3.2. All games were also tested on their most recent public versions, to make sure everything was on an even playing field.

3DMark isn’t exactly reflective of real-world gaming performance, but it does give a good idea of what potential performance looks like for the RTX 5060 Ti, especially as the driver matures. In Speed Way, the RTX 5060 Ti gets 4,209 points, compared to 3,276 from the 4060 Ti, making for a 28% improvement. Steel Nomad and Port Royal see similar improvements, with a 25% and 28% lead respectively in favor of the RTX 5060 Ti.

In Call of Duty Black Ops 6, the numbers don’t fare quite as well. At 1080p with the Extreme preset and DLSS set to Quality, the RTX 5060 Ti gets a comfortable 123 fps. That’s just a 6% lead over the RTX 4060 Ti, but it’s still 35% faster than the RTX 3060 Ti with the same settings.

Luckily, the 5060 Ti does much better in Cyberpunk 2077, which is still one of the most demanding PC games out there right now. At 1080p with the Ray Tracing Ultra preset and DLSS set to Quality, the RTX 5060 Ti gets 90 fps, compared to 74 from the RTX 4060 Ti and 64 from the 3060 Ti. That’s a 21% and a 41% improvement, respectively.

I test Metro Exodus: Enhanced Edition without upscaling to get a picture of what raw ray tracing performance looks like. And the RTX 5060 Ti gets an average of 66 fps at 1080p with the Extreme preset, compared to 58 fps from the 4060 Ti. That’s just a 14% improvement, so not exactly the RTX 5060 Ti’s strongest showing.

Red Dead Redemption 2 is getting old, but it still gives a good portrayal of Vulkan performance, especially when everything is cranked up to max. But even on max settings at 1080p, the RTX 5060 Ti gets 101 fps, compared to 91 fps from the RTX 4060 Ti and 77 fps from the 3060 Ti. That’s an 11% and 31% lead, respectively.

Total War: Warhammer 3 doesn’t have any kind of ray tracing or upscaling technology, so it’s great for seeing how a graphics card handles pure rasterization, and the RTX 5060 Ti gets 137 fps, compared to 106 from the RTX 5060 Ti. That’s one of the best results in the 5060 Ti’s favor, and is extremely promising for anyone who doesn’t really care about fancy ray tracing or upscaling effects in their games.

Assassins Creed Shadows is the newest game in the test suite, and is extremely demanding, especially when all the ray tracing effects are enabled – so of course I enabled them. With everything enabled, the RTX 5060 Ti gets just 45 fps at 1080p. That looks bad, right? Well, the RTX 4060 Ti actually fares much worse, getting just 32 fps with the same settings. That makes for a huge 40% improvement in a game that just came out.

In Black Myth: Wukong, ray tracing is kind of always enabled at some level, but I test the game on the Cinematic preset, without enabling full ray tracing. This makes it one of the most demanding games in my suite, with the RTX 5060 Ti getting just 59 fps at 1080p, falling just short of the 60 fps gold standard. However, it does make for a substantial uplift over the 4060 Ti’s 47 fps.

Jackie Thomas is the Hardware and Buying Guides Editor at IGN and the PC components queen. You can follow her @Jackiecobra