Report
Rubin NVL72 reportedly hits 862 tokens per second per user on GLM5.3 744B agentic workloads
SemiAnalysis says Nvidia increased Rubin interactivity by close to 2x in less than eight days using low-latency specialized kernels.
TLDR
SemiAnalysis reported that Rubin NVL72 reached 862 tokens per second per user on GLM5.3 744B agentic workloads. It also said Nvidia increased Rubin interactivity by close to 2x in less than eight days through low-latency specialized kernels.
Combined views
—
2 Sources, first seen ago
