Nvidia reportedly makes its first Rubin NVL72 Kimi K3 submission

SemiAnalysis says it can reach over 850 tokens per second per user, even at 2.8 trillion parameters.

8 Sources, 1h ago, first seen 1h ago

TLDR

SemiAnalysis says Nvidia submitted its first Rubin NVL72 Kimi K3 submission. It claims the system can reach over 850 tokens per second per user at 2.8 trillion parameters, with over three times higher throughput on vLLM. The October 27 post also said Rubin Software 1.0 general availability was coming soon.

Today's Rank

#15

Combined views

—

8 Sources, first seen 1h ago

Nvidia reportedly makes its first Rubin NVL72 Kimi K3 submission

TLDR

Combined views

8 Sources

Combined views

Sentiment

Summary

Sentiment

Summary

Sentiment

Summary

Sentiment

Summary

8 Sources

8 Sources

8 Sources