Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and H200 (NVIDIA Hopper) on Kimi K3 2.8T. 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 →
GB200 NVL72 posts 4452 tok/s/chip for $0.12 per million tokens at 51 tok/s/user on Kimi K3 2.8T; H200 posts 70 tok/s/chip for $4.85. GB200 NVL72 is 4079% cheaper per token; GB200 NVL72 delivers 6272% more tok/s/chip.
Throughput at 100 tok/s/user on Kimi K3 2.8T: GB200 NVL72 hits 2395 tok/s/chip, H200 hits 70. Per-million costs land at $0.22 and $4.85 respectively. GB200 NVL72 is 2149% cheaper per token; GB200 NVL72 delivers 3328% more tok/s/chip.
GB200 NVL72 / H200 on Kimi K3 2.8T at 149 tok/s/user: 1074 / 70 tok/s/chip, $0.48 / $4.85 per million tokens. GB200 NVL72 is 908% cheaper per token; GB200 NVL72 delivers 1437% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL72:4452.1H200:69.9 | GB200 NVL72:2395.5H200:69.9 | GB200 NVL72:1073.6H200:69.9 |
| Cost ($/M tok) | GB200 NVL72:$0.116H200:$4.850 | GB200 NVL72:$0.216H200:$4.850 | GB200 NVL72:$0.481H200:$4.850 |
| tok/s/MW | GB200 NVL72:2380813H200:51001 | GB200 NVL72:1280990H200:51001 | GB200 NVL72:574116H200:51001 |
| Concurrency | GB200 NVL72:~16H200:~1 | GB200 NVL72:~16H200:~1 | GB200 NVL72:~2H200:~1 |
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.
