DeepSeek V4.1 Flash 552B · Chip comparison
DeepSeek V4.1 Flash 552B — H200 vs Vera Rubin NVL72
Head-to-head AI inference benchmark comparison of H200 (NVIDIA Hopper) and Vera Rubin NVL72 (NVIDIA Vera Rubin) on DeepSeek V4.1 Flash 552B. 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 →
H200 posts 10,388 tok/s/chip for $0.033 per million tokens at 221 tok/s/user on DeepSeek V4.1 Flash 552B; Vera Rubin NVL72 posts 2,818,035 tok/s/chip for $0.000. Vera Rubin NVL72 is 99% cheaper per token; Vera Rubin NVL72 delivers 271× the tok/s/chip.
Throughput at 262 tok/s/user on DeepSeek V4.1 Flash 552B: H200 hits 6,278 tok/s/chip, Vera Rubin NVL72 hits 2,621,324. Costs per million tokens land at $0.054 and $0.000 respectively. Vera Rubin NVL72 is 99% cheaper per token; Vera Rubin NVL72 delivers 418× the tok/s/chip.
H200 / Vera Rubin NVL72 on DeepSeek V4.1 Flash 552B at 302 tok/s/user: 5,746 / 2,385,872 tok/s/chip, $0.059 / $0.000 per million tokens. Vera Rubin NVL72 is 99% cheaper per token; Vera Rubin NVL72 delivers 415× the tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | H20010,388Vera Rubin NVL722,818,035 | H2006,278Vera Rubin NVL722,621,324 | H2005,746Vera Rubin NVL722,385,872 |
| Cost ($/M tok) | H200$0.033Vera Rubin NVL72$0.000 | H200$0.054Vera Rubin NVL72$0.000 | H200$0.059Vera Rubin NVL72$0.000 |
| tok/s/MW | H2007,582,359Vera Rubin NVL72853,950,130 | H2004,582,764Vera Rubin NVL72794,340,588 | H2004,194,222Vera Rubin NVL72722,991,659 |
| Concurrency | H200~5Vera Rubin NVL72~12,898 | H200~2Vera Rubin NVL72~11,485 | H200~1Vera Rubin NVL72~11,316 |