Nebius achieves NVIDIA Exemplar Cloud on NVIDIA HGX B300 for training workloads
Nebius achieves NVIDIA Exemplar Cloud on NVIDIA HGX B300 for training workloads
Nebius has achieved NVIDIA Exemplar Cloud on NVIDIA HGX B300 for training workloads. NVIDIA Exemplar Cloud recognizes cloud platforms that meet NVIDIA’s performance requirements across a defined suite of AI workloads. NVIDIA validated our production infrastructure using its Performance Benchmarking recipes, evaluating training performance against its reference architecture.
This achievement follows our Exemplar Cloud validations on NVIDIA H200 and NVIDIA HGX GB300 NVL72.
How NVIDIA Exemplar Cloud validates cloud performance
Cloud providers run NVIDIA’s standardized benchmarking recipes, measure results against its reference performance, and submit those results to NVIDIA for review. Nebius and NVIDIA engineers tested our production infrastructure, reviewed the results, and refined the platform configuration.
Each GPU platform requires its own validation. Achieving Exemplar on HGX B300 extends our work with the NVIDIABlackwell Ultra platform available on Nebius AI Cloud. This result builds on the engineering approach behind our Exemplar validation on NVIDIA H200 and NVIDIA GB300 NVL72 achievements: designing, operating, and optimizing hardware and cloud software together for customer workloads.
What repeat external validation means for training on Nebius
Exemplar is a strong signal that Nebius’ HGX B300 infrastructure is ready for the most demanding production training workloads, backed by NVIDIA’s validation of training performance. For customers planning demanding training runs, the external validation provides greater confidence in the platform, supported by measured workload performance.
Beyond Exemplar, separate evaluations provide additional evidence of our infrastructure’s performance and reliability. In MLPerf® Training 6.0, Nebius posted the fastest single-node NVIDIA HGX B300 times for Llama-3.1-8B and GPT-OSS 20B in the closed division. SemiAnalysis recently rated Nebius Platinum in ClusterMAX™ 3.0, its highest tier, following testing of training and inference workloads, storage, and failure recovery on live NVIDIA HGX B300 and GB300 NVL72 clusters.
Train on Nebius today
NVIDIA HGX B300 clusters are available on Nebius AI Cloud today. Discover NVIDIA HGX B300 on Nebius, or talk to an expert about your requirements.
Explore Nebius Token Factory
See also
Nebius has achieved NVIDIA Exemplar Cloud status on NVIDIA GB300 NVL72 for training, meeting the 95% performance threshold across NVIDIA’s reference benchmarking recipes. In this post, we explain what this validation means in practice and how it fits into Nebius’s broader work with NVIDIA on building production-grade AI infrastructure.
The Q3 2026 release of Nebius AI Cloud is here. Cloud Interconnect opens a private, MACsec-encrypted path between Nebius and your own networks; spot pricing puts idle GPU capacity on the market at demand-set prices; and Serverless AI now runs Python jobs with no containers to build. Here’s what shipped, what changed, and why it matters.
SemiAnalysis has named Nebius a Platinum AI cloud provider, the top tier of its ClusterMAX™ 3.0 rating. The rating is based on live testing of production clusters: burn-ins, storage under concurrent load, and automated failure recovery on real NVIDIA GB300 and NVIDIA HGX B300 hardware. This post covers what the testing found and why it matters for teams choosing where to run production AI.