AI Agents, StandardNVIDIA Vera Rubin’s MLPerf Results: Training Gains and an Agentic Efficiency Claim November 17, 2026