LLM Digest

Story

vllm_blog · Nov 12, 2026 · news

Source brief

DeepSeek-V4.1-Flash on Vera Rubin NVL72: 58x Throughput vs. GB300 NVL72 at 387 tokens/s

vllm.aiNov 12, 2026
original source linked

In brief

On one NVIDIA Vera Rubin NVL72 rack, vLLM serves DeepSeek-V4.1-Flash on SemiAnalysis AgentX at 1.7M tokens/s per GPU at a P90 of 387 tokens/s per user, 58x GB300 NVL72 at the same interactivity.

Feed lens

agent

Read the original at vllm.ai →Open in live feed Read that day’s brief

DeepSeek-V4.1-Flash on Vera Rubin NVL72: 58x Throughput vs. GB300 NVL72 at 387 tokens/s

Earlier in this thread 4 items

Serving DeepSeek-V4.1-Flash at Scale: vLLM Matches DeepInfer and Scales to GB300 NVL72

NVIDIA Vera Rubin NVL72 Delivers Leading Performance in MLPerf Inference v6.1 Debut

H3-Image Day-0 Support: 4K in 0.87s with Vera Rubin

NVIDIA Vera CPUs double throughput for Modal Sandboxes