Qwen3.6-35B-A3B on Jetson AGX Thor — NVFP4, vLLM vision-language inference benchmark

Open this configuration in the VLM Inference Benchmark Explorer

Model	Qwen3.6-35B-A3B
Parameters	35B
Device	Jetson AGX Thor
Quantization	NVFP4
Image size	720p
Images per camera	1
Max Cameras (response ≤ 3 s)	8
Engine	vLLM

Cameras	Response (s)	TPS (tok/s)	Target
1	0.42	72.1	≤3 s
2	0.65	59.5	≤3 s
4	1.13	46.0	≤3 s
8	1.90	32.2	≤3 s
16	3.14	20.3	>3 s
32	4.06	10.4	>3 s
64	6.64	6.2	>3 s
96	9.19	4.3	>3 s
104	9.93	3.9	>3 s
112	—	—	Failed

Images per request	Response (s)	TPS (tok/s)	Target
1	0.42	72.1	≤3 s
2	0.70	71.5	≤3 s
4	1.36	71.6	≤3 s
8	2.61	69.3	≤3 s
16	5.32	67.1	>3 s
32	11.3	62.2	>3 s
64	25.6	54.9	>3 s
128	62.7	44.3	>3 s
256	177.0	31.5	>3 s
512	—	—	Failed

Qwen3.6-35B-A3B (35B parameters), served in NVFP4 format with vLLM on Jetson AGX Thor, starts answering a single camera that sends one 720p image after 0.42 s, then writes 72.1 tok/s.

With 256 images per camera at 720p, the answer starts after 177.03 s. With one 2K image, after 1.96 s.

At the default target of an answer starting within 3 s (720p, one image per camera), it keeps up with 8 cameras at once.