Record · corpus.blog/posts/01a0f8d0-64db-722e-a9f1-3c33da3e4cec

GB300 NVL72 vs GB200 NVL72 Inference Performance & Perf per Dollar - on DeepSeek-V4-Pro 1.6T: Up to 2.83x Throughput

inferencex.semianalysis.com · published 27 May 2026

A record is what was published and who pointed at it. The text of the post is not held here: read it at the source, or at its Wayback capture.

Read the post at inferencex.semianalysis.com Wayback capture

0blogs citing

First cited: —
Most recent: —
Rank this month: not ranked
External links: 4
Words captured: 1,636

Cited by

No blog we hold has cited this post yet.

Links from this post

Link

Host

https://semianalysis.com/ai-cloud-tco-model/

semianalysis.com

https://api-docs.deepseek.com/news/news260424

api-docs.deepseek.com

https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro

huggingface.co

https://github.com/SemiAnalysisAI/InferenceX/actions/runs/26306422380

github.com

Elsewhere on inferencex.semianalysis.com (16)

https://inferencex.semianalysis.com/blog?tag=benchmark

inferencex.semianalysis.com

https://inferencex.semianalysis.com/blog?tag=gpu

inferencex.semianalysis.com

https://inferencex.semianalysis.com/blog?tag=inference

inferencex.semianalysis.com

https://inferencex.semianalysis.com/blog?tag=deepseek

inferencex.semianalysis.com

https://inferencex.semianalysis.com/blog?tag=nvidia

inferencex.semianalysis.com

https://inferencex.semianalysis.com/blog?tag=gb300

inferencex.semianalysis.com

https://inferencex.semianalysis.com/blog?tag=gb200

inferencex.semianalysis.com

https://inferencex.semianalysis.com/blog?tag=nvl72

inferencex.semianalysis.com

https://inferencex.semianalysis.com/blog?tag=vllm

inferencex.semianalysis.com

https://inferencex.semianalysis.com/blog?tag=dynamo

inferencex.semianalysis.com

https://inferencex.semianalysis.com/blog?tag=wide-ep

inferencex.semianalysis.com

https://inferencex.semianalysis.com/blog?tag=disagg

inferencex.semianalysis.com

https://inferencex.semianalysis.com/gpu-specs

inferencex.semianalysis.com

https://inferencex.semianalysis.com/inference?g_rundate=2026-05-22&g_runid=26306422380&i_active=gb200_dynamo-vllm%2Cgb300_dynamo-vllm&g_model=DeepSeek-V4-Pro&i_linelabel=1

inferencex.semianalysis.com

GB200 NVL72 vs B200 on DeepSeek R1 670B: Up to 4.4x Throughput per GPU at 125 tok/s/user Read at the source ↗

inferencex.semianalysis.com

Vera Rubin NVL72 vs GB200 NVL72? Inference TCO & Architecture Analysis Read at the source ↗

inferencex.semianalysis.com