---
type: "raw_signal"
id: "vllm-adds-day-0-support-for-nvidia-vera-d8d51c"
slug: "vllm-adds-day-0-support-for-nvidia-vera-d8d51c"
title: "vLLM adds day-0 support for NVIDIA Vera Rubin NVL72 via nightly cu134 containers."
description: "vLLM says its runtime now supports NVIDIA Vera Rubin NVL72, with daily container builds and support for models including DeepSeek, Moonshot AI, Z.ai, and MiniMax."
retrieval_nugget: "vLLM says its runtime now supports NVIDIA Vera Rubin NVL72, with daily container builds and support for models including DeepSeek, Moonshot AI, Z.ai, and MiniMax."
observed_at: "2026-10-10"
record_date: "2026-10-10"
date_kind: "observed_at"
why_it_matters: "The preview describes use of CUDA 13.4 locality domains to place MoE weight shards near the compute units that consume them. vLLM reports an early result of 7.8× throughput per GPU versus GB200 NVL72 on AgentX at matched interactivity. The post characterizes the result as early; it should not be read as a general performance guarantee. https://vllm.ai/blog/2026-10-09-vera-rubin-preview"
novelty: "notable"
verification_level: "source-linked"
signal_type: "release"
source_platform: "vllm.ai"
topics: []
entities: ["vLLM","NVIDIA","Vera Rubin NVL72","Blackwell","FlashInfer","CUDA 13.4","DeepSeek","MiniMax"]
related_patterns: ["harness-architecture-outlives-model-choice"]
source_url: "https://vllm.ai/blog/2026-10-09-vera-rubin-preview"
source_urls: ["https://vllm.ai/blog/2026-10-09-vera-rubin-preview"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "signal:vllm-adds-day-0-support-for-nvidia-vera-d8d51c"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/signals/vllm-adds-day-0-support-for-nvidia-vera-d8d51c/","markdown":"https://newruntime.com/signals/vllm-adds-day-0-support-for-nvidia-vera-d8d51c.md","json":"https://newruntime.com/signals/vllm-adds-day-0-support-for-nvidia-vera-d8d51c.json"}
---

# vLLM adds day-0 support for NVIDIA Vera Rubin NVL72 via nightly cu134 containers.

## Retrieval answer

vLLM says its runtime now supports NVIDIA Vera Rubin NVL72, with daily container builds and support for models including DeepSeek, Moonshot AI, Z.ai, and MiniMax.

vLLM says its runtime now supports NVIDIA Vera Rubin NVL72, with daily container builds and support for models including DeepSeek, Moonshot AI, Z.ai, and MiniMax. The project says Blackwell kernels are compatible with Rubin and that Rubin-tuned attention, GEMM, and MoE kernels are available through FlashInfer 0.7.0.

The preview describes use of CUDA 13.4 locality domains to place MoE weight shards near the compute units that consume them. vLLM reports an early result of 7.8× throughput per GPU versus GB200 NVL72 on AgentX at matched interactivity. The post characterizes the result as early; it should not be read as a general performance guarantee. https://vllm.ai/blog/2026-10-09-vera-rubin-preview

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- collection: [Raw Signals](https://newruntime.com/signals.json): Open every source-linked raw signal.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)