At CES 2025 in Las Vegas, NVIDIA presented not only the new GeForce RTX 50 series GPUs, but also many other innovations, including DLSS 4, RTX AI and RTX Neural Shaders.
Powered by NVIDIA Blackwell architecture, fifth-generation Tensor Cores and fourth-generation RT Cores, the GeForce RTX 50 Series is led by the flagship GeForce RTX 5090. According to CEO Jensen Huang, this is the fastest GeForce RTX GPU in Always. The board features 92 billion transients, providing over 3.352 trillion AI operations per second (TOPS) of computing power. Blackwell and DLSS 4 architectural innovations make the GeForce RTX 5090 GPU twice as powerful as the previous GeForce RTX 4090.
“Blackwell, the AI engine, is finally available for gamers, developers and creators,” Huang said. “A fusion of AI-driven neural rendering and ray tracing, Blackwell represents the most significant innovation since we introduced programmable shading 25 years ago.”
Below we report the technical specifications of the NVIDIA GeForce RTX 50 series GPUs.
As regards prices and availability, however, the GeForce RTX 5090 and the GeForce RTX 5080 will be available starting from January 30th for €2,389 and €1,199 respectively. The GeForce RTX 5070 Ti and GeForce RTX 5070 will be available starting in February for €899 and €659 respectively. As always, the Founder’s Editions will be available for purchase on the NVIDIA website and through selected retailers.
During the presentation of the new features, the first details of the DLSS 4 technology were released. This uses Multi Frame Generation for the first time to increase the frame rate by using AI to generate up to three frames per rendered frame. It works in unison with the suite of DLSS technologies to improve performance up to 8x compared to traditional rendering, while maintaining high responsiveness thanks to NVIDIA Reflex technology. DLSS 4 also introduces the first graphical application of the transformer model architecture. Transformer-based DLSS Ray Reconstruction and Super Resolution models use double the parameters and 4x more computing power to deliver better stability, reduced ghosting, greater detail and enhanced anti-aliasing game scenes. DLSS 4 will be supported by GeForce RTX 50 Series in over 75 games and applications available on day 1. NVIDIA Reflex 2 introduces Frame Warp, an innovative technique to reduce latency in gaming, improving the rendered frame based on the last mouse input just before that is sent to the display. Reflex 2 can reduce latency by up to 75%.
Along with the GeForce RTX 50 Series, NVIDIA also introduced RTX Neural Shaders which brings small AI networks into programmable shaders, giving access to cinematic-quality assets, lighting and more in real-time gaming. Rendering game characters is one of the most complicated tasks in terms of real-time graphics, as we tend to notice even the smallest errors or artifacts in digital humans. RTX Neural Faces takes as input a simple rasterized face and 3D pose data, and uses generative AI to render a temporarily stable digital face, in high quality, in real time. RTX Neural Faces is accompanied by new RTX ray-traced hair and skin technologies. Along with the new RTX Mega Geometry, which allows you to increase the number of ray-traced triangles in a scene by up to 100x.
NVIDIA then unveiled a suite of new NVIDIA ACE technologies that enable game characters to perceive, plan and act like human players. ACE-powered autonomous characters have been integrated into KRAFTON’s PUBG: BATTLEGROUNDS and life sim game InZOI, as well as Wemade Next’s MIR5. In PUBG, NVIDIA ACE-powered companions plan and execute strategic actions, dynamically collaborating with players to ensure survival. InZOI features Smart Zoi characters who self-adjust behaviors based on life goals and in-game events. In MIR5, raid bosses guided by large language models (LLMs) adapt tactics based on player behavior, creating more dynamic and challenging encounters.
To top it off, GeForce RTX 50 Series GPUs power creative workflows. RTX 50 Series GPUs are the first consumer GPUs to support FP4 precision, increasing AI image generation performance by twofold for models like FLUX and allowing generative AI models to run locally with a smaller memory footprint, compared to previous generation hardware. The NVIDIA Broadcast app adds two AI-powered beta features for livestreamers: Studio Voice, which improves microphone audio, and Virtual Key light, which illuminates faces for clearer streams. Streamlabs is introducing the Intelligent Streaming Assistant, powered by NVIDIA ACE and Inworld AI, which acts as a cohost, producer and technical assistant to enhance livestreams.
