VLM Inference Benchmark Explorer
NVIDIA-Nemotron-3.5-Bolt-74B-A9B on DGX Spark — NVFP4, vLLM vision-language inference benchmark
Open this configuration in the VLM Inference Benchmark Explorer
| Model | NVIDIA-Nemotron-3.5-Bolt-74B-A9B |
|---|---|
| Parameters | 74B |
| Device | DGX Spark |
| 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 | 0.66 | 21.4 | ≤3 s |
| 2 | 0.78 | 19.4 | ≤3 s |
| 4 | 1.16 | 15.4 | ≤3 s |
| 8 | 2.11 | 11.8 | ≤3 s |
| 16 | 3.48 | 8.3 | >3 s |
| 32 | 5.62 | 5.5 | >3 s |
| 64 | 9.51 | 3.3 | >3 s |
| 72 | — | — | Failed |
| Images per request | Response (s) | TPS (tok/s) | Target |
|---|---|---|---|
| 1 | 0.66 | 21.4 | ≤3 s |
| 2 | 0.98 | 21.4 | ≤3 s |
| 4 | 2.04 | 21.5 | ≤3 s |
| 8 | 3.97 | 21.9 | >3 s |
| 16 | 7.74 | 21.8 | >3 s |
| 32 | — | — | Failed |
NVIDIA-Nemotron-3.5-Bolt-74B-A9B (74B parameters), served in NVFP4 format with vLLM on DGX Spark, starts answering a single camera that sends one 720p image after 0.66 s, then writes 21.4 tok/s.
With 16 images per camera at 720p, the answer starts after 7.74 s. With one 2K image, after 2.03 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.
NVIDIA-Nemotron-3.5-Bolt-74B-A9B (74B parameters), served in NVFP4 format with vLLM on DGX Spark, starts answering a single camera that sends one 720p image after 0.66 s, then writes 21.4 tok/s.
With 16 images per camera at 720p, the answer starts after 7.74 s. With one 2K image, after 2.03 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.