Jensen Huang took the stage at CES on Monday to share the latest from NVIDIA, and while the presentation was more of a reminder of the technologies the company has been working on in recent years, there were a few notable announcements.
NVIDIA announced Alpamayo, a family of open-source reasoning models designed to guide autonomous vehicles in difficult driving situations. The centerpiece of this release is Alpamayo 1, a 10 billion-parameter thought chain system, according to NVIDIA, capable of approaching driving as a human would. The model works by breaking down unexpected driving situations into a smaller set of problems before finding the safest path. At each step, the model can explain its reasoning.
A sister model called AlpaSim allows developers to complete closed-loop training for driving scenarios rarely encountered in real life. Huang said the Mercedes Benz CLA 2025 will be the first vehicle to ship with NVIDIA’s entire AV stack, including Alpamayo. “Our vision is that one day every car, every truck will be autonomous,” Huang said.
Following Alpamayo’s announcements, a pair of BD-1 droids from Star Wars Jedi: Fallen Order joined Huang on stage. We saw one join the executive at last year’s CES. After that, Huang turned to Vera Rubin. Nvidia first announced GPU architecture in 2024and now the company has started production of a super computer that uses the new technology. A Vera processor has 88 custom Olympus cores and 1.5 TB of system memory for a total of 227 billion transistors. Meanwhile, a Rubin GPU has 336 billion transistors. Each Vera Rubin supercomputer has a pair of both components.
Following the presentation, NVIDIA held a separate briefing during which announced DLSS 4.5 and G-Sync Pulsar. The latest version of NVIDIA’s upscaling technology has been built on a second-generation transformer model, which should reduce ghosting and flare, leading to a more stable image, even when there is a lot of movement on the screen. As part of DLSS 4.5, NVIDIA is also adding support for multi-frame and dynamic 6x generation. Both features will arrive in spring. The first allows a 50-series GPU to generate five frames for every traditionally rendered frame. The idea here is to allow a powerful GPU like the RTX5090 to saturate a 4K, 240 Hz screen with as many images as possible. Dynamic image generation, on the other hand, is exactly what it sounds like. DLSS 4.5 can dynamically adapt the number of images generated to fit the scenario. In demanding scenes, your 50 Series GPU will generate more frames, while reducing frames in less hectic scenes so it only computes what it needs.
As for Pulsar G-Syncthis is the latest improvement in NVIDIA’s flicker reduction technology. By pulsing a display’s backlight, NVIDIA claims it can deliver perceived motion clarity relative to 1,000Hz, leading to greater clarity. These same displays will also come with the ability to automatically adjust their brightness and color temperature to ambient lighting conditions. Pre-orders for the first batch of G-Sync Pulsar displays will open on January 7th.
Updated 01/06/26 9:30 a.m.: Added information about DLSS 4.5 and G-SYNC Pulsar.
This article was originally published on Engadget at https://www.engadget.com/ai/everything-nvidia-announced-at-ces-2026-225653684.html?src=rss
