VLM Inference Benchmark Explorer
Cosmos-Reason2-2B on Jetson AGX Orin — Q8_0, llama.cpp vision-language inference benchmark
Open this configuration in the VLM Inference Benchmark Explorer
| Model | Cosmos-Reason2-2B |
|---|---|
| Parameters | 2.4B |
| Device | Jetson AGX Orin |
| Quantization | Q8_0 |
| Image size | 720p |
| Images per camera | 1 |
| Max Cameras (response ≤ 3 s) | 4 |
| Engine | llama.cpp |
| Cameras | Response (s) | TPS (tok/s) | Status |
|---|---|---|---|
| 1 | 0.95 | 57.8 | PASS |
| 2 | 1.09 | 30.2 | PASS |
| 4 | 2.13 | 16.5 | PASS |
| 8 | 4.95 | 23.3 | FAIL |
Cosmos-Reason2-2B (2.4B parameters), served in Q8_0 format with llama.cpp on Jetson AGX Orin, starts answering a single camera that sends one 720p image after 0.95 s, then writes 57.8 tok/s.
With 5 images per camera at 720p, the answer starts after 4.45 s. With one 2K image, after 7.07 s.
At the default target of an answer starting within 3 s (720p, one image per camera), it keeps up with 4 cameras at once.
Cosmos-Reason2-2B (2.4B parameters), served in Q8_0 format with llama.cpp on Jetson AGX Orin, starts answering a single camera that sends one 720p image after 0.95 s, then writes 57.8 tok/s.
With 5 images per camera at 720p, the answer starts after 4.45 s. With one 2K image, after 7.07 s.
At the default target of an answer starting within 3 s (720p, one image per camera), it keeps up with 4 cameras at once.