Benchmarking organization MLCommons released MLPerf Inference v6.1 results, introducing two new tests, including agentic inference, and recording its highest submitting-organization count.

MLCommons released results for MLPerf Inference v6.1, its inference benchmark suite. The announcement says the release set a record for the number of submitting organizations, but the supplied excerpt does not give that count.

The suite introduces two new tests aligned with emerging AI deployment patterns, including agentic inference. The available description does not name the second test or provide the complete workloads and measurement rules.

Related NVIDIA coverage reports the Vera Rubin NVL72 system’s debut in this benchmark release. A separate NVIDIA Developer Blog title reports that TensorRT Edge-LLM completes the MLPerf edge agentic benchmark 6.4× faster on Jetson AGX Thor, but the supplied excerpt does not identify the comparison baseline, so the multiplier cannot be treated as a general cross-platform performance claim.