VLM Inference Benchmark Explorer

Qwen3.5-27B on DGX Spark — FP8, vLLM vision-language inference benchmark

Open this configuration in the VLM Inference Benchmark Explorer

ModelQwen3.5-27B
Parameters27B
DeviceDGX Spark
QuantizationFP8
Image size720p
Images per camera1
Max Cameras (TTFT ≤ 3000 ms)8
EnginevLLM
CamerasTTFT (ms)TPS (tok/s)Target
15458.1≤3000 ms
29618.1≤3000 ms
417527.7≤3000 ms
826527.0≤3000 ms
1642765.8>3000 ms
3266984.2>3000 ms
4891013.3>3000 ms
56102832.9>3000 ms
60——Failed
Images per requestTTFT (ms)TPS (tok/s)Target
15458.1≤3000 ms
210418.0≤3000 ms
420278.0≤3000 ms
841248.0>3000 ms
1685207.8>3000 ms
32180237.6>3000 ms
64404007.2>3000 ms
128978366.5>3000 ms
256——Failed

Qwen3.5-27B (27B parameters), served in FP8 format with vLLM on DGX Spark, starts answering a single camera that sends one 720p image after 545 ms (time to first token, TTFT), then writes 8.1 tok/s.

With 128 images per camera at 720p, the answer starts after 97,836 ms. With one 2K image, after 2,394 ms.

At the default target of a mean TTFT within 3,000 ms (720p, one image per camera), it keeps up with 8 cameras at once.