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
| Model | Qwen3.5-27B |
|---|---|
| Parameters | 27B |
| Device | DGX Spark |
| Quantization | FP8 |
| Image size | 720p |
| Images per camera | 1 |
| Max Cameras (TTFT ≤ 3000 ms) | 8 |
| Engine | vLLM |
| Cameras | TTFT (ms) | TPS (tok/s) | Target |
|---|---|---|---|
| 1 | 545 | 8.1 | ≤3000 ms |
| 2 | 961 | 8.1 | ≤3000 ms |
| 4 | 1752 | 7.7 | ≤3000 ms |
| 8 | 2652 | 7.0 | ≤3000 ms |
| 16 | 4276 | 5.8 | >3000 ms |
| 32 | 6698 | 4.2 | >3000 ms |
| 48 | 9101 | 3.3 | >3000 ms |
| 56 | 10283 | 2.9 | >3000 ms |
| 60 | — | — | Failed |
| Images per request | TTFT (ms) | TPS (tok/s) | Target |
|---|---|---|---|
| 1 | 545 | 8.1 | ≤3000 ms |
| 2 | 1041 | 8.0 | ≤3000 ms |
| 4 | 2027 | 8.0 | ≤3000 ms |
| 8 | 4124 | 8.0 | >3000 ms |
| 16 | 8520 | 7.8 | >3000 ms |
| 32 | 18023 | 7.6 | >3000 ms |
| 64 | 40400 | 7.2 | >3000 ms |
| 128 | 97836 | 6.5 | >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.
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.