Nvidia Vera Rubin NVL72 delivers 30x more work per watt than GB300 on agentic workloads

nordicinst · x · 2026-08-24

Nvidia released performance data for Vera Rubin NVL72 systems, showing up to 30x higher throughput per megawatt and 35x lower token costs compared to GB300 NVL72 on agentic workloads. Measured using SemiAnalysis AgentX real-world coding trajectories, the results highlight significant efficiency gains as AI agents demand exponentially more tokens for long-context tasks.

Related event: NVIDIA's Vera Rubin Delivers 30x Gains Over GB300 in Agent Workloads(2 posts)→

Original post →

More from Infra

Infra channel →