MLPerf Training 6.0 fastest single-node HGX B300 results
Company: Nebius
The claim, verbatim
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.
Source (primary)
Nebius achieves NVIDIA Exemplar Cloud on NVIDIA HGX B300 for training workloads - Nebius (-, news_article)
View cached copy (2026-10-11)Live source ↗Archive.org ↗
Quote: “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.”
How we checked this
Checked on October 15, 2026. The cited source supports every part of this claim.
The post makes this exact statement about Nebius's MLPerf Training 6.0 results.
Confirmed in the source:
- In MLPerf Training 6.0, Nebius posted the fastest single-node NVIDIA HGX B300 time for Llama-3.1-8B in the closed division
- In MLPerf Training 6.0, Nebius posted the fastest single-node NVIDIA HGX B300 time for GPT-OSS 20B in the closed division
What we did: Read our cached copy of the publisher (https://nebius.com/blog/posts/nebius-achieves-nvidia-exemplar-cloud-on-nvidia-hgx-b300-for-training) in full (5,691 characters, retrieved October 11, 2026) and checked each assertion in the claim against it.
Additional evidence
confirms Nebius achieves NVIDIA Exemplar Cloud on NVIDIA HGX B300 for training workloads - Nebius
Quote: “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.”