Head-to-head AI inference benchmark comparison of GB200 NVL72 (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 — 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 →
At 127 tok/s/user interactivity on DeepSeekv4 Pro 0813 1.6T, GB200 NVL72 delivers 27183 tok/s/chip at $0.02 per million tokens; Vera Rubin NVL72 delivers 258818 tok/s/chip at $0.00. Vera Rubin NVL72 is 391% cheaper per token; Vera Rubin NVL72 delivers 852% more tok/s/chip at this point.
GB200 NVL72 posts 10412 tok/s/chip for $0.05 per million tokens at 181 tok/s/user on DeepSeekv4 Pro 0813 1.6T; Vera Rubin NVL72 posts 153620 tok/s/chip for $0.01. Vera Rubin NVL72 is 660% cheaper per token; Vera Rubin NVL72 delivers 1375% more tok/s/chip.
Throughput at 234 tok/s/user on DeepSeekv4 Pro 0813 1.6T: GB200 NVL72 hits 4278 tok/s/chip, Vera Rubin NVL72 hits 5513. Per-million costs land at $0.12 and $0.18 respectively. GB200 NVL72 is 51% cheaper per token; Vera Rubin NVL72 delivers 29% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL72:27182.8Vera Rubin NVL72:258817.8 | GB200 NVL72:10412.5Vera Rubin NVL72:153619.6 | GB200 NVL72:4278.0Vera Rubin NVL72:5513.1 |
| Cost ($/M tok) | GB200 NVL72:$0.019Vera Rubin NVL72:$0.004 | GB200 NVL72:$0.050Vera Rubin NVL72:$0.007 | GB200 NVL72:$0.121Vera Rubin NVL72:$0.182 |
| tok/s/MW | GB200 NVL72:14536279Vera Rubin NVL72:78429629 | GB200 NVL72:5568169Vera Rubin NVL72:46551396 | GB200 NVL72:2287686Vera Rubin NVL72:1670632 |
| Concurrency | GB200 NVL72:~223Vera Rubin NVL72:~2771 | GB200 NVL72:~31Vera Rubin NVL72:~1458 | GB200 NVL72:~5Vera Rubin NVL72:~23 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
