Oriel — AI & technology worth noticing, every day.

NVIDIA Sweeps MLPerf Training and Endpoints as Vera Rubin Debuts With Leading Performance

Web page · https://blogs.nvidia.com/blog/mlperf-training-endpoints-vera-rubin

In its MLPerf benchmarks debut, with results released today, the NVIDIA Vera Rubin platform delivers leading performance across training, post-training and agentic inference. NVIDIA swept every MLPerf Training v6.1 benchmark and, in its first MLPerf Endpoints submission, delivered more than 10x generational gains in throughput per watt for agentic AI — measuring how much agentic […]

First seen by Oriel on .

AI summary

NVIDIA reports Vera Rubin’s MLPerf debut, claiming leading training and agentic-inference performance, including a complete MLPerf Training v6.1 sweep and over 10x throughput-per-watt gains for agentic AI.

  • Vera Rubin was the only platform with results on every MLPerf Training v6.1 benchmark, covering language models, image generation, recommender systems and reinforcement-learning post-training.
  • Vera Rubin NVL72 delivered up to 2x more performance per GPU than GB300 NVL72, according to the reported results.
  • On DeepSeek-V3, NVL72 scaled from 256 to 1,024 GPUs at about 97% efficiency, reducing training time from 16 minutes to approximately four.
  • For Kimi K3, Vera Rubin NVL72 reported over 10x GB300’s throughput per watt at a shared minimum interactivity of 50 tokens per second per user.
  • The Kimi K3 Vera Rubin result is provisional and pending MLCommons Association verification, while the GB300 comparison was verified.

#vera rubin #mlperf #agentic inference #ai infrastructure

Generated by gpt-5.6-sol. Check the original for details.

Aliases

SourceExternal IDURL
NVIDIA Bloghttps://blogs.nvidia.com/?p=99933https://blogs.nvidia.com/blog/mlperf-training-endpoints-vera-rubin/ Read discussion