Qwen3.5-2B on Jetson AGX Orin 32GB Developer Kit — Q8_0, llama.cpp vision-language inference benchmark

Open this configuration in the VLM Inference Benchmark Explorer

Model	Qwen3.5-2B
Parameters	2.3B
Device	Jetson AGX Orin 32GB Developer Kit
Quantization	Q8_0
Image size	720p
Images per camera	1
Max Cameras (response ≤ 3 s)	4
Engine	llama.cpp

Cameras	Response (s)	TPS (tok/s)	Target
1	0.95	47.6	≤3 s
2	1.43	31.4	≤3 s
4	2.86	34.1	≤3 s
8	8.84	37.0	>3 s

Images per request	Response (s)	TPS (tok/s)	Target
1	0.95	47.6	≤3 s
2	1.80	48.0	≤3 s
4	3.49	47.2	>3 s
8	6.88	45.9	>3 s
16	13.8	44.0	>3 s
32	28.0	40.3	>3 s

Qwen3.5-2B (2.3B parameters), served in Q8_0 format with llama.cpp on Jetson AGX Orin 32GB Developer Kit, starts answering a single camera that sends one 720p image after 0.95 s, then writes 47.6 tok/s.

With 32 images per camera at 720p, the answer starts after 28.01 s. With one 2K image, after 6.99 s.

At the default target of an answer starting within 3 s (720p, one image per camera), it keeps up with 4 cameras at once.