NVIDIA Dynamo Operations Center
Real-time visibility and control of the 8 × B300 GPU cluster infrastructure
Total GPU utilisation
92.4%
▲ 4.8%
Token throughput
2.48M/s
▲ 12.6%
Requests / second
18.7K
▲ 9.2%
Models in catalog
8
0 tenant deployments
Dynamo scheduler
98.7%
▲ 2.3%efficiency
Revenue today
$482,391
▲ 16.8%
Real-time inference metrics
Prefill and decode throughput across all clusters
Token distribution by model
- gpt-4o23%
- claude-3.518%
- llama-3.1-70b15%
- nemotron-ultra13%
- others31%
GPU cluster overview
Four clusters · 8 × B300 each
Cluster-A
Healthy94%
GPU utilisation
Memory 89%62°C24 models612K t/s
Cluster-B
Healthy91%
GPU utilisation
Memory 82%58°C22 models548K t/s
Cluster-C
Healthy93%
GPU utilisation
Memory 87%64°C26 models587K t/s
Cluster-D
Healthy90%
GPU utilisation
Memory 84%60°C20 models521K t/s
Live routing decisions
Last 60 seconds
| Request | Intent | Model | Latency |
|---|---|---|---|
| req-7843 | Coding | deepseek-r1 | 162 ms |
| req-7842 | Research | gpt-4o | 542 ms |
| req-7841 | FAQ | llama-3.1-70b | 118 ms |
| req-7840 | Agentic | nemotron-ultra-550b | 487 ms |
| req-7839 | Vision | gpt-4o | 623 ms |