VLM Inference Benchmark Explorer
Qwen3-VL-4B-Instruct on Jetson Orin NX — FP8, vLLM vision-language inference benchmark
Open this configuration in the VLM Inference Benchmark Explorer
| Model | Qwen3-VL-4B-Instruct |
|---|---|
| Parameters | 4.4B |
| Device | Jetson Orin NX |
| Quantization | FP8 |
| Image size | 720p |
| Images per camera | 1 |
| Max Cameras (response ≤ 3 s) | 2 |
| Engine | vLLM |
| Cameras | Response (s) | TPS (tok/s) | Status |
|---|---|---|---|
| 1 | 1.67 | 14.2 | PASS |
| 2 | 2.59 | 13.1 | PASS |
| 4 | 4.74 | 13.1 | FAIL |
Qwen3-VL-4B-Instruct (4.4B parameters), served in FP8 format with vLLM on Jetson Orin NX, starts answering a single camera that sends one 720p image after 1.67 s, then writes 14.2 tok/s.
With 3 images per camera at 720p, the answer starts after 4 s. With one 1080p image, after 4.34 s.
At the default target of an answer starting within 3 s (720p, one image per camera), it keeps up with 2 cameras at once.
Qwen3-VL-4B-Instruct (4.4B parameters), served in FP8 format with vLLM on Jetson Orin NX, starts answering a single camera that sends one 720p image after 1.67 s, then writes 14.2 tok/s.
With 3 images per camera at 720p, the answer starts after 4 s. With one 1080p image, after 4.34 s.
At the default target of an answer starting within 3 s (720p, one image per camera), it keeps up with 2 cameras at once.