DeepSeek R1 — GB200 NVL72 vs Vera Rubin NVL72
Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and Vera Rubin NVL72 (NVIDIA Vera Rubin) on DeepSeek R1. Latency, throughput, and cost across LLM workloads. Use the chart controls below to switch sequences, precisions, and metrics — same interactions as the main inference chart.
Setting 114 tok/s/user as the target on DeepSeek R1, GB200 NVL72 produces 9917 tok/s/chip ($0.05 per million tokens) and Vera Rubin NVL72 produces 0 ($0.00). Comparable data is unavailable.
At 176 tok/s/user interactivity on DeepSeek R1, GB200 NVL72 delivers 2899 tok/s/chip at $0.18 per million tokens; Vera Rubin NVL72 delivers 0 tok/s/chip at $0.00. Comparable data is unavailable at this point.
GB200 NVL72 posts 896 tok/s/chip for $0.58 per million tokens at 238 tok/s/user on DeepSeek R1; Vera Rubin NVL72 posts 0 tok/s/chip for $0.00. Comparable data is unavailable. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL72:9917.2Vera Rubin NVL72:0.0 | GB200 NVL72:2898.6Vera Rubin NVL72:0.0 | GB200 NVL72:896.5Vera Rubin NVL72:0.0 |
| Cost ($/M tok) | GB200 NVL72:$0.052Vera Rubin NVL72:$0.000 | GB200 NVL72:$0.178Vera Rubin NVL72:$0.000 | GB200 NVL72:$0.576Vera Rubin NVL72:$0.000 |
| tok/s/MW | GB200 NVL72:5303313Vera Rubin NVL72:0 | GB200 NVL72:1550046Vera Rubin NVL72:0 | GB200 NVL72:479388Vera Rubin NVL72:0 |
| Concurrency | GB200 NVL72:~772Vera Rubin NVL72:~400 | GB200 NVL72:~103Vera Rubin NVL72:~400 | GB200 NVL72:~16Vera Rubin NVL72:~400 |