MiniCPM-V-4.6 on Jetson Orin NX 16GB — Q8_0, llama.cpp vision-language inference benchmark

Open this configuration in the VLM Inference Benchmark Explorer

Model	MiniCPM-V-4.6
Parameters	1.3B
Device	Jetson Orin NX 16GB
Quantization	Q8_0
Image size	720p
Images per camera	1
Max Cameras (response ≤ 3 s)	1
Engine	llama.cpp

Cameras	Response (s)	TPS (tok/s)	Target
1	1.94	43.0	≤3 s
2	3.44	44.4	>3 s
4	9.51	49.6	>3 s
8	19.5	44.6	>3 s

Images per request	Response (s)	TPS (tok/s)	Target
1	1.94	43.0	≤3 s
2	3.85	42.9	>3 s
4	7.64	42.4	>3 s
8	15.2	41.4	>3 s
16	30.6	39.5	>3 s
32	62.0	36.2	>3 s

MiniCPM-V-4.6 (1.3B parameters), served in Q8_0 format with llama.cpp on Jetson Orin NX 16GB, starts answering a single camera that sends one 720p image after 1.94 s, then writes 43 tok/s.

With 32 images per camera at 720p, the answer starts after 62.04 s. With one 2K image, after 2.62 s.

At the default target of an answer starting within 3 s (720p, one image per camera), it keeps up with one camera at a time.