NVIDIA Vera Rubin NVL72 hits up to 5.7x GB300 throughput per MW on agent benchmark
ArtificialAnlys · x · 2026-10-16
Artificial Analysis published first Vera Rubin NVL72 results on its AA-AgentPerf agentic inference benchmark: up to 5.7x GB300 NVL72's total throughput per MW on DeepSeek V4 Pro (5.4x on DeepSeek V4.1 Flash, 3.0x on GLM-5.2, 2.8x on Kimi K3). The rack-scale system pairs 72 Rubin GPUs with 36 Vera CPUs; each Rubin GPU keeps Blackwell Ultra's 288 GB memory but delivers 2.4x memory bandwidth, 1.7x NVLink bandwidth and 2.3x dense NVFP4 compute. All results were submitted by NVIDIA running TensorRT-LLM and vLLM with NVIDIA Dynamo.
Related event: Vera Rubin NVL72 hits 5.7x GB300 throughput per MW in first tests(3 posts)→
More from Infra
- TRL reproduces Modal's mixed-precision async RL: plain GRPO collapses, SC+MIS holds 0.72 — QGallouedec · 2026-10-16
- Hugging Face scales Indic pretraining with 2.8T tokens and an open data pipeline — Hugging Face Blog · 2026-10-16
- 14 open-source LLM observability tools to know in 2026, from Langfuse to Laminar — TheTuringPost · 2026-10-16
- Local AI is a routing problem, not a privacy feature: where each task should run — bigdata · 2026-10-16
- Moving agent work from GPUs to CPUs for durable, predictable actions — cramforce · 2026-10-16
- First TPUv7 agentic benchmark on vLLM claims 1.42x perf-per-dollar over GB300 NVL72 — vllm_project · 2026-10-16