StockDuty ← Dashboard Nov 18, 2026 03:05 PM ET
SignalTech/AI

NVIDIA Vera Rubin NVL72 Leads Agentic Inference Performance on MLPerf Endpoints v1.0

NVIDIA Vera Rubin NVL72 leads agentic inference performance on MLPerf Endpoints v1.0.

What happened: NVIDIA published provisional MLPerf Endpoints v1.0 results showing Vera Rubin NVL72 outperforming GB300 NVL72 on throughput per kilowatt for agentic inference workloads. Key numbers: Kimi K3 10x throughput per kW ratio, DeepSeek-V4.1-Flash 9x, Qwen3.6-35B-A3B 5x. Vera Rubin Kimi K3 run: 41,414.2 tokens/sec at 100.89 kW, 50.4 tokens/sec per user. Why it matters: The results compare serving capacity at a shared minimum interactivity of 50 tokens per second per user. Vera Rubin results are provisional and pending MLCommons Association verification and peer review.
NVIDIA published provisional MLPerf Endpoints v1.0 results showing Vera Rubin NVL72 outperforming GB300 NVL72 on throughput per kilowatt for agentic inference workloads.
The results compare serving capacity at a shared minimum interactivity of 50 tokens per second per user. Vera Rubin results are provisional and pending MLCommons Association verification and peer review.
Sources