Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and Vera Rubin NVL72 (NVIDIA Vera Rubin) on DeepSeekv4 Pro 0813 1.6T. Latency, throughput, and cost across LLM workloads. Use the chart controls below to switch sequences, precisions, and metrics β same interactions as the main inference chart.
AgentX replays real coding-agent sessions rather than fixed-length prompts, so context grows turn over turn and most of each request is served from cache instead of being recomputed. That turns the comparison into a systems question: KV transfer between nodes, prefix-aware routing, and cache capacity all move the curve alongside raw chip throughput. Fixed-sequence workloads stay the clean baseline for kernel and silicon performance, so the two scenarios answer different questions about the same hardware. Learn more about AgentX β
At 129 tok/s/user interactivity on DeepSeekv4 Pro 0813 1.6T, B200 delivers 22787 tok/s/chip at $0.02 per million tokens; Vera Rubin NVL72 delivers 256644 tok/s/chip at $0.00. Vera Rubin NVL72 is 440% cheaper per token; Vera Rubin NVL72 delivers 1026% more tok/s/chip at this point.
B200 posts 9035 tok/s/chip for $0.05 per million tokens at 185 tok/s/user on DeepSeekv4 Pro 0813 1.6T; Vera Rubin NVL72 posts 142100 tok/s/chip for $0.01. Vera Rubin NVL72 is 654% cheaper per token; Vera Rubin NVL72 delivers 1473% more tok/s/chip.
Throughput at 240 tok/s/user on DeepSeekv4 Pro 0813 1.6T: B200 hits 3989 tok/s/chip, Vera Rubin NVL72 hits 4903. Per-million costs land at $0.12 and $0.20 respectively. B200 is 70% cheaper per token; Vera Rubin NVL72 delivers 23% more tok/s/chip. (Numbers reflect the default agentic-traces Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B200:22786.6Vera Rubin NVL72:256643.8 | B200:9034.6Vera Rubin NVL72:142100.1 | B200:3989.4Vera Rubin NVL72:4903.3 |
| Cost ($/M tok) | B200:$0.021Vera Rubin NVL72:$0.004 | B200:$0.053Vera Rubin NVL72:$0.007 | B200:$0.120Vera Rubin NVL72:$0.205 |
| tok/s/MW | B200:13325510Vera Rubin NVL72:77770858 | B200:5283404Vera Rubin NVL72:43060633 | B200:2332981Vera Rubin NVL72:1485860 |
| Concurrency | B200:~56Vera Rubin NVL72:~2749 | B200:~12Vera Rubin NVL72:~1288 | B200:~4Vera Rubin NVL72:~21 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
