NVIDIA unveils Rubin platform and open models at CES
NVIDIA opened CES in Las Vegas with a six‑chip Rubin AI platform and new open models for autonomy.
NVIDIA founder and CEO Jensen Huang took the keynote stage to unveil Rubin and introduce Alpamayo, the company said.
Rubin is NVIDIA’s first extreme‑codesigned, six‑chip AI platform and is now in full production, the company said. Alpamayo is an open reasoning model family for autonomous vehicle development, the company added.
Rubin platform components include Rubin GPUs with 50 petaflops of NVFP4 inference; Vera CPUs; NVLink 6; Spectrum‑X Ethernet Photonics; ConnectX‑9 SuperNICs; and BlueField‑4 DPUs, the chipmaker detailed.
The company also introduced NVIDIA Inference Context Memory Storage Platform, an AI‑native KV‑cache tier it said boosts long‑context inference with 5x higher tokens per second, 5x better performance per TCO dollar and 5x better power efficiency.
“Computing has been fundamentally reshaped as a result of accelerated computing, as a result of artificial intelligence,” Huang said.
He said the Rubin platform is intended to reduce the cost of training and inference and speed model development by delivering AI tokens at one‑tenth the cost.
Related reading
- NVIDIA's Drive automation debuts in Mercedes‑Benz CLA
- GeForce NOW kicks off January with 14 new games as Blackwell upgrade boosts cloud performance
- Liquid Reply joins Linux Foundation–backed Agentic AI Foundation
The firm said its open models are trained on NVIDIA supercomputers and span six domains, including Clara for healthcare, Cosmos for robotics and Alpamayo for autonomous driving.
Huang also announced Alpamayo R1 — the first open, reasoning VLA model for autonomous driving — and AlpaSim, a fully open simulation blueprint for high‑fidelity AV testing. He said the first passenger car featuring Alpamayo on the NVIDIA DRIVE full‑stack platform will be on the roads soon in the all‑new Mercedes‑Benz CLA, with AI‑defined driving coming to the U.S. this year.
The Recap
- NVIDIA unveiled the Rubin platform and open models at CES.
- Rubin GPUs deliver 50 petaflops of NVFP4 inference performance.
- AI-defined driving coming to the U.S. this year.
