Along with three other GPUs, the RTX 5070 was finally announced during CES 2025, and it is currently scheduled to be released later this month. Ever since Nvidia’s official announcement, many have been wondering how the mid-range card performs against other options in the market, including some GPUs from Nvidia itself. One of the most obvious comparisons has to be against its predecessor, the RTX 4070, as everyone would like to know what improvements it offers.

While real-world benchmarks of the RTX 5070 are not here, Nvidia has shared the specifications of the card. Thanks to this, we have some information that can help us compare the cards on paper while we wait for real-world benchmarks to become available. However, due to strict embargos, we and many others won’t be able to reveal the benchmarks for a few more weeks, so you’ll need to wait for that.

Here’s how the RTX 5070 and RTX 4070 stack up against each other on paper so far.

RTX 5070 vs RTX 4070 specs comparison

As you can see in the table above, both GPUs offer the same amount of VRAM, but the 5070 offers a better memory type. On top of that, it also has more cores, better clock speeds, and a higher bandwidth.

Memory

When it comes to the VRAM, the RTX 5070 offers 12 GB of GDDR7 memory with a 192-bit memory interface and a bandwidth of 672.2 GB/s. On the other hand, its predecessor, the RTX 4070 also offers 12 GB VRAM, but GDDR6 memory. While the memory interface is the same, the bandwidth is 504.2 GB/s.

The higher the VRAM of your GPU, the easier it will be for you to adjust and improve the graphical settings in a game, especially at higher resolutions like 4K. However, it’s a bit disappointing to see that the 5070 has the same VRAM as its predecessor. Sure, the better memory type will help it deliver better performance even with the same amount of VRAM, but we were hoping to see at least 16 GB in the 5070.

AMD is also introducing 16 GB VRAM in their upcoming mid-range cards. On the other hand, Intel also recently introduced two budget GPUs with 10 and 12 GB VRAM. So, for mid-range cards, 16 GB should be the standard now.

Clock speeds and cores 

The RTX 5070 has a base clock speed of 2.16 GHz and a boost clock speed of 2.51 GHz. On the other hand, the RTX 4070 has base and boost clock speeds of 1.92 and 2.47 GHz, respectively. This means that thanks to the better speeds, the 5070 can process more data per second than the 4070. 

Aside from that, the RTX 5070 also takes the lead when it comes to the core count, which was already expected. It offers 6,144 cores, while the 4070 offers 5,888 cores. This means that there is a difference of 256 cores between the two. 

Sure, the difference between the two isn’t as big as the RTX 5090 and 4090, which is over 5,000, but it’s still something that will help the 5070 gain a performance advantage over the 4070. On top of that, the card also offers fifth-generation Tensor cores and fourth-generation RT cores, meaning each core on the 5070 is much more powerful and advanced than the ones present on the 4070.

Features 

Nvidia has an upscaling technology called DLSS. This is available for all RTX GPUs, including the 20, 30, and 40-series cards. Right now, we’re using DLSS 3.5, which debuted back in 2020 and has been helping users with getting more FPS in games to this day. However, it will soon be replaced by DLSS 4, a much more advanced version of the upscaling technology.

DLSS 4 was announced during CES 2025 and it will be released alongside two RTX 50-series cards, the RTX 5090 and RTX 5080, at the end of this month. Once it gets released, it will be available to all those who were able to use the previous version.

However, one important feature, Multi Frame Generation, will be exclusive to the RTX 50-series. This feature helps generate three additional frames per traditionally rendered frame, which is massive. In fact, Nvidia currently claims that the RTX 5070 will offer performance similar to the RTX 4090, and we believe that it’s because of DLSS 4 considering the card doesn’t match its specifications.

So, in short, while the RTX 5070 already has better specs than the 4070 on paper, DLSS 4 is something that will put it in an entirely different league. In the video shared by Nvidia below, you can see how DLSS 4 offers over 100 more frames than DLSS 3.5 in Cyberpunk 2077.  

Another feature announced during CES 2025 is Reflex 2. While this won’t be exclusive to RTX 50-series cards, it will debut on them first. Nvidia has not shared when it will become available for previous lineups, but it will happen sometime in the future.

For those wondering, Reflex is Nvidia’s way of reducing latency in games. At the moment, we have Reflex 1, which many gamers are using to make their gameplay smoother. In the image below, Nvidia reveals how Reflex 1 reduced the latency in The Finals from 56 ms to 27 ms, and Reflex 2 slashed it further to just 14 ms. This will be useful especially for competitive gamers, as it eliminates lag and stuttering. 

So, with DLSS 4 and Reflex 2, the RTX 5070 will offer much better performance than the RTX 4070. It might not have been able to achieve the same with just its specifications, but the two new technologies should put it way ahead of its predecessor. 

RTX 5070 vs RTX 4070 – price comparison

According to Nvidia, the RTX 5070 will launch with a price tag of $549. However, keep in mind that this will be for the base model, as the third-party variants will most likely carry a higher price tag – somewhere between $650 and $700.

On the other hand, the RTX 4070 was launched with a $599 price tag. These days, you can find it either above or below this price, depending on the variant. Once the 5070 hits the market, the price of the 4070 will most likely drop.

Early impressions

Based on the specifications shared by Nvidia, the RTX 5070 offers specs better than its predecessor, the RTX 4070. It offers a better memory type, more CUDA cores, higher bandwidth, and a better architecture. On top of that, the card will also benefit greatly from the Multi Frame Generation feature of DLSS 4, which will help it deliver performance similar to the RTX 4090, at least that’s what Nvidia says.

However, we’re a bit disappointed to see only 12 GB VRAM in the 5070. Competitors like AMD are introducing 16 GB VRAM in their upcoming mid-range options, while one of Intel’s budget options has 12 GB VRAM. With the 5070, we were hoping to see at least 16 GB of memory.

In any case, once the card hits the market, we’ll make sure to update this guide with our in-depth review and real-world benchmarks.