What is NVIDIA Vera Rubin CPX Platform?
Vera Rubin CPX is a new class of NVIDIA GPU purpose-built for massive-context inference, capable of processing million-token inputs for complex generative AI and long-sequence reasoning. Each Rubin CPX GPU delivers up to 50 petaFLOPS of NVFP4 compute with 128GB of GDDR7 memory, optimized for cost and efficiency.
In the Vera Rubin platform, Rubin CPX handles the ‘context’ or prefill phase of inference, while standard Rubin GPUs focus on decode phase. This disaggregated approach accelerates large-scale reasoning and video-generation workloads with greater throughput and lower energy use.
