Jetson Thor T5000: 90 benchmark rows, 17% measured.

Method v1.0 · dataset 2026-09-07 · verified 2026-09-07 · last updated September 2026

15 of 90 rows are measured and published (class C); the rest are EdgeAIStack-derived (class D). 10 model families. Precisions: FP16 (37), FP8 (21), INT8 (32). Runtimes: tensorrt (90). Tasks: classification (5), detection (37), pose (24), segmentation (24). Jetson Thor T5000 platform page.

Family × precision: best value

The fastest matching row per family and precision (fps, or tokens/s for language models), with its evidence class. Click a cell for the full breakdown.

Family	FP16	FP8	INT8
MobileNetV2	3640 fps D	—	4004 fps D
ResNet	2100 fps D	2940 fps D	2520 fps D
YOLO11	719 fps D	1150 fps D	657 fps D
YOLO11	629 fps D	689 fps D	575 fps D
YOLO11	483 fps D	530 fps D	441 fps D
YOLO12	719 fps C	—	—
YOLO26	719.4 fps C	—	657.9 fps C
YOLOv8	537 fps D	859 fps D	491 fps D
YOLOv8	508 fps D	813 fps D	464 fps D
YOLOv8	371 fps D	594 fps D	339 fps D

Query this hardware live.

Filter by model, precision or runtime and get the matching rows with a permanent link.

OPEN BENCHMARK EXPLORER →

Jetson Thor T5000: 90 benchmark rows, 17% measured.

Family × precision: best value

Every model on Jetson Thor T5000

Links

Query this hardware live.