Skip to main content

Leaked Nvidia GeForce RTX 3070 Ti could topple charts with double the memory

Less than 24 hours after Nvidia unveiled its performance chart-topping GeForce RTX 3000 series graphics cards — including the RTX 3070 — a new round of leaks suggest that the GPU maker is already working on an even more powerful midrange refresh. The GeForce RTX 3070 Ti is now tipped to come with even more video memory, which should further boost performance.

Recommended Videos

The unannounced GeForce RTX 3070 Ti was recently spotted online, according to a report from PC Gamer, and the card was listed in several SKUs of Lenovo’s upcoming Legion T7 gaming systems for Russia that was posted by Videocardz. Lenovo’s listing suggests that the card would have 16GB of GDDR6 video RAM, double what’s available on the non-Ti variant.

Image source: Videocardz Image used with permission by copyright holder

The Legion T7 gaming desktops are also listed with Intel Core i7-10700K or i9-10900K processors and up to 32GB of RAM. It’s unclear at this time how accurate Lenovo’s posting is, or if the manufacturer is relying on unverified specifications from partner Nvidia as placeholders for what these systems will actually contain when they ship.

Get your weekly teardown of the tech behind PC gaming
Check your inbox!

Additionally, Twitter user @_rogame stated that MSI would also make a version of the RTX 3070 Ti with 16GB of memory. At this point, it remains unclear if Nvidia will rely on the “Ti” branding for the card when it launches, or if it will use the new “Super” branding.

Given that Nvidia CEO Jensen Huang previously claimed that the vanilla non-Ti variant of the RTX 3070 already comes with enough performance to topple the current GeForce RTX 2080 Ti on performance, we can expect the RTX 3070 Ti to be a very strong, competitive contender if and when it launches. Aside from using different architectures — the RTX 2000 series was based on Nvidia’s Turing architecture, whereas the RTX 3000 series uses the more modern Ampere architecture — a major difference between the RTX 3070 and the RTX 2080 Ti was memory configuration.

Whereas RTX 2080 Ti shipped with 11GB of RAM, the RTX 3070 only came with 8GB. A Ti version of this midrange card would help it best the prior generation flagship on paper, and upgraded specifications will help Nvidia stay competitive in this segment against upcoming offerings from rival AMD.

Image used with permission by copyright holder

For reference, the GeForce RTX 3070 comes with a boost speed of 1,730 MHz and with 5,888 CUDA cores. That card is coming in October and Nvidia is pricing the RTX 3070 at just $499. It’s speculated that the 16GB RTX 3070 Ti would retain the 256-bit memory bus and stick with the slower GDDR6 RAM rather than the faster GDDR6X memory on the RTX 3080 and RTX 3090. The RTX 3080 also jumps to a faster 320-bit memory bus, while the RTX 3090 goes to a 384-bit bus.

If Nvidia does in fact launch the Ti variant, it’s unclear how the company will price the new card. The flagship RTX 3080 was announced at $699, so Nvidia could price the RTX 3070 Ti somewhere between the cost of the RTX 3070 and the RTX 3080, or it could ultimately be priced to replace the RTX 3070 when it launches.

Chuong Nguyen
Silicon Valley-based technology reporter and Giants baseball fan who splits his time between Northern California and Southern…
Nvidia’s RTX 40-series is coming to an end
Three RTX 4080 cards sitting on a pink background.

Out with the old, in with the new. According to Board Channels, Nvidia has now halted production for nearly all of its best graphics cards as it shifts focus to the RTX 50-series. Only one GPU remains in production, and some of the cards that are the most in demand are no longer being produced.

Nvidia hasn't officially announced that it's sunsetting the RTX 40-series, but we've been hearing more and more reports that imply that might be the case. The RTX 4090 was among the first cards to go out of production, and the discontinuation appears to have immediately affected the markets. Nvidia's behemoth flagship was hard to come by at the best of times, and now, as no more new units are being produced, it's safe to assume that this situation won't improve. The cheapest RTX 4090 I could find on Amazon costs nearly $2,000, but you can still snag one for .

Read more
GeForce Now introduces a 100-hour monthly playtime cap
nvidia geforce now ultimate halfway across country 1

Nvidia’s GeForce Now is undergoing significant changes. Its Priority membership has been rebranded as the "Performance" tier, bringing an enhanced gaming experience at the same price. Performance members can now stream games at up to 1440p resolution and in ultrawide formats, with settings saved across sessions. This upgrade aligns the Performance tier closer to Ultimate, the highest tier, which supports 4K streaming and high refresh rates through GeForce RTX 4080-powered servers.

One of the most notable changes coming in 2025 is the introduction of a 100-hour monthly playtime cap for both Performance and Ultimate members, aimed at managing Nvidia’s cloud infrastructure. Nvidia also mentions in its blog post that this limit will affect only a small percentage of its player base, with over 94% of users typically playing well within this timeframe. To offer flexibility, Nvidia will allow up to 15 unused hours to roll over to the following month. Users who reach their limit can purchase extra playtime, with additional 15-hour blocks available at $2.99 for Performance members and $5.99 for Ultimate members.

Read more
Nvidia’s RTX 50-series may launch ‘soon,’ whatever that means
Nvidia CEO Jensen Huang with an RTX 4090 graphics card.

As we inch closer to the expected release date of Nvidia's RTX 50-series, the number of leaks is growing by the minute. Today, a reputable leaker weighed in on when we might see the RTX 50-series join the ranks of the best graphics cards. Could Blackwell make an appearance sooner than currently expected? It's certainly possible, but who even knows at this point?

The information comes from kopite7kimi, who, in typical tipster fashion, dropped a vague message on X (formerly Twitter) and then left without answering any questions. However, at this point in the GPU release cycle, even one vague sentence is enough to send the internet for a spin, which is what's happening in the reply section of Kopite's tweet.

Read more