DeepSeek R1 — B300 vs Vera Rubin NVL72
Head-to-head AI inference benchmark comparison of B300 (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.
Throughput at 119 tok/s/user on DeepSeek R1: B300 hits 4426 tok/s/chip, Vera Rubin NVL72 hits 0. Per-million costs land at $0.14 and $0.00 respectively. Comparable data is unavailable.
B300 / Vera Rubin NVL72 on DeepSeek R1 at 185 tok/s/user: 1477 / 0 tok/s/chip, $0.43 / $0.00 per million tokens. Comparable data is unavailable.
Toward the upper edge of the 52–318 tok/s/user interactivity band, at 252 tok/s/user on DeepSeek R1: B300 runs 849 tok/s/chip at $0.74/M tokens, Vera Rubin NVL72 runs 0 at $0.00/M. Comparable data is unavailable. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:4426.3Vera Rubin NVL72:0.0 | B300:1476.6Vera Rubin NVL72:0.0 | B300:849.4Vera Rubin NVL72:0.0 |
| Cost ($/M tok) | B300:$0.142Vera Rubin NVL72:$0.000 | B300:$0.425Vera Rubin NVL72:$0.000 | B300:$0.739Vera Rubin NVL72:$0.000 |
| tok/s/MW | B300:2329647Vera Rubin NVL72:0 | B300:777161Vera Rubin NVL72:0 | B300:447058Vera Rubin NVL72:0 |
| Concurrency | B300:~85Vera Rubin NVL72:~400 | B300:~41Vera Rubin NVL72:~400 | B300:~28Vera Rubin NVL72:~400 |