Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and H100 (NVIDIA Hopper) on Muse Spark Open 816B. 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
Throughput at 84 tok/s/user on Muse Spark Open 816B: GB200 NVL72 hits 25162 tok/s/chip, H100 hits 1316. Per-million costs land at $0.02 and $0.25 respectively. GB200 NVL72 is 1103% cheaper per token; GB200 NVL72 delivers 1812% more tok/s/chip.
GB200 NVL72 / H100 on Muse Spark Open 816B at 112 tok/s/user: 19270 / 1316 tok/s/chip, $0.03 / $0.25 per million tokens. GB200 NVL72 is 821% cheaper per token; GB200 NVL72 delivers 1364% more tok/s/chip.
Toward the upper edge of the 57β166 tok/s/user interactivity band, at 139 tok/s/user on Muse Spark Open 816B: GB200 NVL72 runs 16901 tok/s/chip at $0.03/M tokens, H100 runs 1316 at $0.25/M. GB200 NVL72 is 708% cheaper per token; GB200 NVL72 delivers 1184% more tok/s/chip. (Numbers reflect the default agentic-traces Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL72:25162.4H100:1315.8 | GB200 NVL72:19270.5H100:1315.8 | GB200 NVL72:16901.2H100:1315.8 |
| Cost ($/M tok) | GB200 NVL72:$0.021H100:$0.247 | GB200 NVL72:$0.027H100:$0.247 | GB200 NVL72:$0.031H100:$0.247 |
| tok/s/MW | GB200 NVL72:13455804H100:960474 | GB200 NVL72:10305078H100:960474 | GB200 NVL72:9038092H100:960474 |
| Concurrency | GB200 NVL72:~19H100:~1 | GB200 NVL72:~14H100:~1 | GB200 NVL72:~12H100:~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.
