Head-to-head AI inference benchmark comparison of H200 (NVIDIA Hopper) 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 133 tok/s/user on DeepSeekv4 Pro 0813 1.6T, Vera Rubin NVL72 delivers 251970 tok/s/chip at $0.00 per million tokens; H200 hasn't been benchmarked at this target.
Vera Rubin NVL72 hits 120764 tok/s/chip for $0.01 per million tokens at 192 tok/s/user on DeepSeekv4 Pro 0813 1.6T. No H200 data at this operating point.
Vera Rubin NVL72: 3963 tok/s/chip, $0.25 per million tokens at 251 tok/s/user on DeepSeekv4 Pro 0813 1.6T. H200 is unmeasured here. (Numbers reflect the default agentic-traces Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | H200:βVera Rubin NVL72:251970.0 | H200:βVera Rubin NVL72:120763.8 | H200:βVera Rubin NVL72:3962.8 |
| Cost ($/M tok) | H200:βVera Rubin NVL72:$0.004 | H200:βVera Rubin NVL72:$0.008 | H200:βVera Rubin NVL72:$0.253 |
| tok/s/MW | H200:βVera Rubin NVL72:76354553 | H200:βVera Rubin NVL72:36595081 | H200:βVera Rubin NVL72:1200851 |
| Concurrency | H200:βVera Rubin NVL72:~2699 | H200:βVera Rubin NVL72:~1037 | H200:βVera Rubin NVL72:~17 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
