NVIDIA Vera Rubin Sweeps MLPerf Debut, Delivers 10x Per-Watt Gain in Agentic Inference
NVIDIA Blog · rss · 2026-11-17
In its MLPerf debut, NVIDIA's Vera Rubin platform swept every MLPerf Training v6.1 benchmark—the only platform with results across the board—and delivered over 10x generational gains in throughput per watt for agentic inference in its first MLPerf Endpoints submission. Vera Rubin NVL72 hits up to 2x per-GPU performance vs GB300 NVL72; on DeepSeek-V3 it scaled 256→1,024 GPUs at 97% efficiency, cutting training from 16 to 4 minutes. Provisional Kimi K3 results show 10x+ per-watt agentic throughput at equal interactivity. Twenty partner organizations submitted results.
More from Infra
- Microsoft's USB4 RDMA lets mini PCs pool memory to run Qwen3-235B — JFPuget · 2026-11-22
- LTX 2.5 keeps crashing on AMD 9070 XT with driver timeouts while MiniMax H3 runs fine — ZestycloseReading504 · 2026-11-22
- FT: AI server suppliers raise prices up to 45% months after deals are signed — nordicinst · 2026-11-22
- China Installed 59% of New Industrial Robots in 2026 Report, ~354,000 Units — ingliguori · 2026-11-22
- MLX over Vulkan hits 124 tok/s on RTX 5060 Ti, beating llama.cpp CUDA by 25% — SumitGup · 2026-11-22
- Intruder swapped Prime Intellect's hosted DeepSeek for an abliterated copy for 53 hours — mawcopolow · 2026-11-22