NVIDIA Vera Rubin Platform Achieves New MLPerf Benchmark Records
In short: NVIDIA has released the initial MLPerf Training v6.1 and Endpoints v1.0 results for its new Vera Rubin architecture. The system demonstrated significant leaps in both training speed and power efficiency for agent-based AI workloads.
This summary was generated automatically by AI from NVIDIA's publication. It is our own text, not a copy of the original — facts, figures and quotes belong to the source, linked above and below.
What changed?
- 1Vera Rubin NVL72 provides double the training speed per GPU compared to GB300 NVL72 systems.
- 2Efficiency in agentic AI tasks increased by over 10x in throughput per watt.
- 3Achieved 97% scaling efficiency when utilizing 1,024 GPUs for training benchmarks.
| Parameter | Before | Now |
|---|---|---|
| Training performance per GPU | GB300 NVL72 | Up to 2x improvement with Vera Rubin NVL72 |
| Agentic AI throughput per watt | GB300 NVL72 | >10x improvement with Vera Rubin NVL72 |
Why it matters
These results indicate a major advancement in the hardware foundations required for faster model fine-tuning and the deployment of complex, multi-step AI agents. Enhanced throughput per watt allows for higher query concurrency and more responsive real-time interactions in large-scale contact center applications.
What it means for AI agents and contact centers
As your infrastructure scales to handle massive concurrent voice streams and agentic tasks, these efficiency gains suggest lower costs for high-volume inference. Monitor how these hardware advancements impact the latency and scalability of your AI-driven customer service deployments.
Sources
- NVIDIAOfficialPrimary sourceOriginal article →„NVIDIA Sweeps MLPerf Training and Endpoints as Vera Rubin Debuts With Leading Performance“16 Nov 2026, 18:00Licence: robots.txt Content-Signal: ai-input=yes · our summary (content changed)
- NVIDIA DeveloperOfficialOriginal article →„NVIDIA Vera Rubin NVL72 Achieves up to 2x Blackwell Ultra Performance in MLPerf Training v6.1“16 Nov 2026, 18:00
- NVIDIA DeveloperOfficialOriginal article →„NVIDIA Vera Rubin NVL72 Leads Agentic Inference Performance on MLPerf Endpoints v1.0“16 Nov 2026, 18:00
- Published by source
- 16 Nov 2026, 18:00
- Found by our system
- 16 Nov 2026, 18:04
- Summary generated
- 16 Nov 2026, 18:05
This article was written by AI from the original source. Facts, numbers and prices come from the source; missing values are marked “Not specified”. Legal notice, copyright and privacy