📢 First ever on-silicon NVIDIA Vera Rubin performance measured on how agents actually run.
⚡ Up to 30x more throughput per megawatt and up to 35x lower token cost than GB300 NVL72.Agentic sessions are nothing like chat or summarization workloads. Context grows across hundreds of steps and can reach hundreds of thousands of tokens. NVIDIA measured Vera Rubin performance on @SemiAnalysis_ AgentX workload consisting of real-world agentic coding trajectories using DeepSeek V4 Pro model.Source: NVIDIA_X
