nvidia

NVIDIA Jetson AGX Orin

Flagship robotics module — runs full VLA + perception stacks alongside SLAM.

top models on this chip

Object detection

Sorted by latency p50 · matched-pair runs on NVIDIA Jetson AGX Orin

Object detection

#	Model	Quant	Latency p50	Throughput	Acc.	Power
1	NVIDIA-YOLO27-Orin	MIXED	6.02 ms	166 FPS	99.6%	48.8 W
2	Meridian Highway-Perception v3	MIXED	7.70 ms	130 FPS	99.0%	41.2 W
3	RT-DETR-Edge	FP16	9.49 ms	105 FPS	99.8%	40.8 W
4	YOLO27-Edge	FP16	9.77 ms	102 FPS	99.4%	49.4 W
5	YOLO-World-Edge	FP16	11.4 ms	88 FPS	99.6%	52.3 W

top models on this chip

Sorted by latency p50 · matched-pair runs on NVIDIA Jetson AGX Orin

Depth

#	Model	Quant	Latency p50	Throughput	Acc.	Power
1	Meridian Depth-Tier1	FP16	8.25 ms	121 FPS	99.6%	40.7 W
2	MiDaS-Distilled-S	FP16	8.49 ms	118 FPS	99.8%	50.9 W
3	Depth-Anything-Edge	MIXED	9.52 ms	105 FPS	99.1%	55.1 W

top models on this chip

Sorted by latency p50 · matched-pair runs on NVIDIA Jetson AGX Orin

Segmentation

#	Model	Quant	Latency p50	Throughput	Acc.	Power
1	Mask2Former-Mobile	FP16	17.9 ms	56 FPS	99.5%	51.5 W
2	SAM-3-Distilled	FP16	21.9 ms	46 FPS	99.7%	49.7 W

top models on this chip

Sorted by latency p50 · matched-pair runs on NVIDIA Jetson AGX Orin

Vision-Language-Action

#	Model	Quant	Latency p50	Throughput	Acc.	Power
1	Meridian LanePolicy-Mini	MIXED	35.9 ms	502 tok/s	99.1%	40.0 W
2	Cosmos-Policy-Lite	MIXED	43.9 ms	410 tok/s	98.6%	45.7 W
3	π0-Distilled	MIXED	48.5 ms	371 tok/s	99.0%	53.1 W
4	V-JEPA-AC-Edge	FP16	65.6 ms	275 tok/s	99.6%	49.2 W
5	RT-2-Mobile	MIXED	83.0 ms	217 tok/s	98.6%	48.2 W

top models on this chip

Sorted by latency p50 · matched-pair runs on NVIDIA Jetson AGX Orin

Multimodal

#	Model	Quant	Latency p50	Throughput	Acc.	Power
1	DINO-v3-Distilled	FP16	26.1 ms	38 FPS	99.7%	55.4 W
2	MobileCLIP-B	FP16	28.7 ms	35 FPS	99.5%	42.2 W
3	Florence-2-Edge	FP16	31.3 ms	32 FPS	99.6%	44.8 W
4	SmolVLM-3B	MIXED	152.7 ms	6.6 FPS	98.8%	50.5 W