MLPerf Inference v6.0 open results — NVIDIA Jetson AGX Thor llama3.1-8b-edge Offline + SingleStream (raw logs)

Fields

id

sources/mlperf-v6-0-nvidia-thor-llama31-8b-edge

URL

https://github.com/mlcommons/inference_results_v6.0/tree/main/open/NVIDIA/results/Jetson-Thor/llama3.1-8b-edge

Publisher

MLCommons / NVIDIA

Kind

benchmark

Published

2026-04-29

Accessed

2026-10-17

notes

Open-division submission. 1x Jetson AGX Thor Developer Kit 128GB, TensorRT 10.8, CUDA 12.8, FP4 weights. Offline log: Tokens per second 350.605, VALID. SingleStream log: 90th-percentile first-token latency and time-to-output-token, VALID. Quoted verbatim from both mlperf_log_summary.txt files 2026-10-17.

Notes

Open-division submission. 1x Jetson AGX Thor Developer Kit 128GB, TensorRT 10.8, CUDA 12.8, FP4 weights. Offline log: Tokens per second 350.605, VALID. SingleStream log: 90th-percentile first-token latency and time-to-output-token, VALID. Quoted verbatim from both mlperf_log_summary.txt files 2026-10-17.

Cross-references

This record references no other records.