BriefsNewsAvailable now
NVIDIA debuts Vera Rubin NVL72 AI platform in MLPerf Training v6.1
The new Vera Rubin NVL72 architecture achieved up to 2x the performance per GPU compared to the GB300 NVL72 baseline in recent industry benchmarks.
Summary
NVIDIA introduced its Vera Rubin NVL72 platform in the latest MLPerf Training v6.1 benchmarks, securing the fastest training times on six of seven tested models. The system demonstrated high scaling efficiency and performance gains across various AI workloads, including reinforcement learning and large-scale pretraining. These systems are currently available through global cloud infrastructure partners.
What changed
- NVIDIA launched the Vera Rubin NVL72 platform for AI training and inference.
- The platform achieved up to 2x performance per GPU compared to the GB300 NVL72 baseline on the DeepSeek-V3 671B benchmark.
- Vera Rubin NVL72 maintained over 97% scaling efficiency when expanding from 256 to 1,024 GPUs.
- The platform supports both TensorRT-LLM and vLLM inference backends.
Availability
Available now“Vera Rubin NVL72 systems are generally available through cloud infrastructure partners worldwide.”
Founder takeaway
If you are managing large-scale AI training workloads, consider the Vera Rubin NVL72 platform for its improved performance per GPU and scaling efficiency. The system is available now through cloud providers.
Source
Researched and drafted with AI from the official sources above, then checked automatically before publishing. Spotted a mistake? Tell us.
Related
More briefs
Google brings the Gemini app to low-memory Android devices
Google re-architected the Gemini app so it can run on eligible Android Go edition phones.
Google AI1 min read
Fireworks AI benchmarks the MiniMax M3.1 Flash AI model
Fireworks AI compares the MiniMax M3.1 Flash AI model against closed alternatives and tests how combining models changes costs and coding completions.
Fireworks AI1 min read
Google adds temporary threads to AI Mode in Search
Google introduced temporary threads in AI Mode so users can ask questions without saving them to their history or affecting recommendations.
Google Search1 min read
One email, every Monday
Briefs are published here through the day. Monday Heist is the email: one useful AI workflow, the files to use it, and the catch. The first issue is being prepared; subscribers get it first. Free, and you can leave in one click.
One email. Every Monday.