- Nvidia’s RTX 5000 collection playing cards have been introduced
- RTX 5090 prices $1,999 / £1,999 / AU$4,039
- 5090 and 5080 go on sale January 30
After months of hypothesis and anticipation, Nvidia lastly lifted the quilt off its newest lineup of shopper graphics playing cards, the Nvidia GeForce RTX 5000 collection, beginning with the flagship RTX 5090.
Nvidia CEO Jensen Huang made the announcement at a packed area at Las Vegas’s Mandalay Bay resort and on line casino, a headline occasion that topped off a day of main reveals from rivals AMD and Intel at CES 2025.
Nevertheless, the Nvidia RTX 5000 collection graphics playing cards have been at all times going to steal the present, it doesn’t matter what anybody else introduced, so Huang naturally had the limelight on Monday evening as he unveiled our first definitive take a look at the RTX 5090 – amongst different new GPUs and fancy AI options.
Meet Nvidia’s new flagship GPU, the Nvidia Tita— I imply the Nvidia RTX 5090
It may not be referred to as the Nvidia Titan RTX, however the RTX 5090 may as effectively be, given the specs on provide and its downright scandalous MSRP of $1,999 (£1,999 / AU$4,039) – not a generational worth bump from the RTX 4090, granted, however nonetheless a ridiculous sum of money for an ostensibly ‘shopper’ graphics card.
With an astounding 92 billion transistors, next-gen Tensor Cores and Ray Tracing Cores, and greater than double the AI processing velocity of the 4090, the RTX 5090 will unquestionably be probably the most highly effective shopper graphics card on the planet, and it will not even be shut.
Pair the GPU specs with 32GB of shiny new GDDR7 VRAM on a large 512-bit reminiscence bus and PCIe 5.0 x16 interface, it has an astonishing 1792 GB/s of reminiscence bandwidth, with a reminiscence velocity upwards of 23.8 Gbps.
RelatedPosts
Given its specs, not solely will this graphics card completely blow via native 4K gaming (with out upscaling) on the highest settings (together with ray tracing), it is arguably the primary actual 8K graphics card given the quantity of VRAM it has and its reminiscence bandwidth, two key specs that enable a graphics card to course of the considerably bigger 8K texture information wanted for gaming at that decision.
In fact, few video games even assist 8K decision, a lot much less have builders and artists successfully losing their time on texture information so massive that solely a uncommon few will ever see them as supposed. However there is not any doubt that if 8K gaming ever turns into a factor, the RTX 5090 might be greater than prepared to fulfill the problem.
In fact, that does not actually handle the truth that that is not a gaming GPU— not anymore, and never at this worth. And if (effectively, when) scalpers get entangled, it will be far worse.
In any case, we referred to as the $1,199 price ticket on the flagship RTX 2080 Ti ‘virtually obscene’ in our evaluate three GPU generations in the past. With no generational worth drop from the already wildly costly RTX 4090, it isn’t exhausting to argue that the RTX 5090 is solely knowledgeable workstation GPU, meant to course of uncooked 4K video streams or render prolonged 3D generated sequences at Pixar or another animation studio. As enjoyable because it may be, this isn’t a graphics card meant to play Wolfenstein 3D.
The Nvidia GeForce RTX 5090 is ready to go on sale for $1,999 on January 30.
The Nvidia GeForce RTX 5080 will launch alongside the 5090
In a transfer that has been telegraphed for some time, the Nvidia GeForce RTX 5080 may also be a part of the primary wave of Nvidia’s new graphics playing cards to hit retailer cabinets later this month.
The RTX 5080 seems to be to be virtually precisely half of the RTX 5090 when it comes to specs. Though Nvidia has but to launch a complete spec sheet for the brand new GPUs – an odd transfer, although possible one designed to keep away from distracting from all the brand new AI options – we all know that the 5080 has 1,801 AI TOPS (trillion operations per second), a bit beneath half the RTX 5090’s 3,352.
It additionally has new GDDR7 VRAM as effectively, with a pool of 16GB on a 256-bit reminiscence bus for 960 GB/s of reminiscence bandwidth – once more, mainly half the specs of the 5090’s VRAM. Its reminiscence velocity is a blazing quick 30 Gbps, which helps make up for the narrower reminiscence bus.
The cardboard will go on January twenty first with an MSRP of $999 (anticipated £999 / AU$2,019), which once more matches the launch worth of the previous-gen RTX 4080.
Nvidia RTX 5070 and RTX 5070 Ti additionally unveiled – however you will have to attend
The Nvidia RTX 5070 and 5070 Ti additionally debuted on Monday evening, possible an effort from Nvidia to reassure players that they have not forgotten concerning the midrange market – particularly with its most important rival AMD refocusing to focus on the finances and midrange house solely with its new GPUs.
Once more, we do not have the breadth of specs we would hoped to see at this level, however we do know that the RTX 5070 Ti gives 1,406 AI TOPS whereas the RTX 5070 has 988 TOPS. As a tough level of comparability, Apple‘s new M4 chip caps out at round 38 TOPS – so a devoted GPU is arguably nonetheless a necessity for critical AI workloads.
Over on the VRAM entrance, the RTX 5070 Ti’s reminiscence profile is almost an identical to the RTX 5080, with 16GB GDDR7, a reminiscence velocity of 28 Gbps on a 256-bit bus, and a reminiscence bandwidth of 896 GB/s, making it greater than prepared for 4K gaming. In the meantime, the RTX 5070 has 12GB of GDDR7 at 672 GB/sec – nonetheless sooner than the RTX 4070, although the identical base quantity of VRAM.
The Nvidia GeForce RTX 5070 and 5070 Ti will go on sale in February – date to be confirmed – with respective retail costs of $549 (£549 / AU$1,509) and $749 (£749 / AU$1,109). That is truly fairly pleasing to see, for the reason that RTX 4070 retailed at $599 – which means we lastly have a generational worth drop from Nvidia.
DLSS 4, Reflex 2, and extra
Along with the brand new {hardware}, Nvidia additionally showcased a number of upgraded AI options debuting with the ‘Blackwell’ RTX 5000 technology of graphics playing cards.
Chief amongst these, after all, was DLSS 4 – the most recent replace to Nvidia’s decision upscaling software program, which permits for higher framerates in-game by rendering the sport at a decrease decision and upscaling it to a goal decision (say, 1080p to 4K) utilizing AI.
DLSS 4 may also characteristic ‘Multi Body Era’, an improved model of the Body Era tech seen within the RTX 4000 technology, which makes use of AI to extrapolate and produce extra frames and ‘insert’ them between odd rendered frames to spice up framerate. Sadly for customers on older GPUs, solely the common DLSS 4 upscaling might be out there on older playing cards; Multi Body-Gen might be unique to RTX 5000 playing cards.
We’re additionally getting Nvidia Reflex 2, a brand new model of the Reflex software program for decreasing enter latency in video games. Reflex 2 will characteristic ‘Body Warp’, which goals to proactively insert generated frames by studying mouse enter earlier than it even reaches the show – this could reportedly scale back enter latency by as a lot as 75%.
RTX 5000 can be bringing AI powers to shader tech with new RTX Neural Shaders. This makes use of small AI networks within the GPU’s programmable shader models to ship ‘film-quality’ shading and lighting in-game. ‘RTX Neural Faces’, together with new RTX tech for hair and pores and skin rendering and animation, guarantees to ship extra reasonable people than ever earlier than.
Lastly, we will anticipate to see extra of Nvidia ACE with this technology of RTX GPUs – the improved AI capabilities of the Blackwell technology imply that initiatives like Nvidia’s (barely creepy) AI NPC tech may be carried out on a wider degree, with ACE-powered characters deliberate to look in a handful of titles together with PUBG: Battlegrounds and InZOI.
