Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and H200 (NVIDIA Hopper) 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 →
At 62 tok/s/user interactivity on DeepSeek V4.1 Flash 552B, GB200 NVL72 delivers 66042 tok/s/chip at $0.01 per million tokens; H200 delivers 13391 tok/s/chip at $0.03. GB200 NVL72 is 223% cheaper per token; GB200 NVL72 delivers 393% more tok/s/chip at this point.
GB200 NVL72 posts 39287 tok/s/chip for $0.01 per million tokens at 115 tok/s/user on DeepSeek V4.1 Flash 552B; H200 posts 7462 tok/s/chip for $0.05. GB200 NVL72 is 245% cheaper per token; GB200 NVL72 delivers 426% more tok/s/chip.
Throughput at 168 tok/s/user on DeepSeek V4.1 Flash 552B: GB200 NVL72 hits 23493 tok/s/chip, H200 hits 4923. Per-million costs land at $0.02 and $0.07 respectively. GB200 NVL72 is 213% cheaper per token; GB200 NVL72 delivers 377% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL72:66042.5H200:13390.8 | GB200 NVL72:39286.5H200:7462.2 | GB200 NVL72:23492.6H200:4922.6 |
| Cost ($/M tok) | GB200 NVL72:$0.008H200:$0.025 | GB200 NVL72:$0.013H200:$0.045 | GB200 NVL72:$0.022H200:$0.069 |
| tok/s/MW | GB200 NVL72:35316829H200:9774328 | GB200 NVL72:21008842H200:5446830 | GB200 NVL72:12562881H200:3593130 |
| Concurrency | GB200 NVL72:~46H200:~19 | GB200 NVL72:~22H200:~9 | GB200 NVL72:~13H200:~6 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
No measurements to plot for this selection. Review the benchmark controls above or adjust quick filters.
