Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and H200 (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 57 tok/s/user on Muse Spark Open 816B: B300 hits 21347 tok/s/chip, H200 hits 6063. Per-million costs land at $0.03 and $0.06 respectively. B300 is 90% cheaper per token; B300 delivers 252% more tok/s/chip.
B300 / H200 on Muse Spark Open 816B at 100 tok/s/user: 21347 / 4493 tok/s/chip, $0.03 / $0.08 per million tokens. B300 is 156% cheaper per token; B300 delivers 375% more tok/s/chip.
Toward the upper edge of the 15β186 tok/s/user interactivity band, at 144 tok/s/user on Muse Spark Open 816B: B300 runs 18037 tok/s/chip at $0.03/M tokens, H200 runs 2777 at $0.12/M. B300 is 251% cheaper per token; B300 delivers 549% more tok/s/chip. (Numbers reflect the default agentic-traces Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:21346.8H200:6062.8 | B300:21346.8H200:4493.2 | B300:18037.2H200:2777.1 |
| Cost ($/M tok) | B300:$0.029H200:$0.056 | B300:$0.029H200:$0.075 | B300:$0.035H200:$0.122 |
| tok/s/MW | B300:11235170H200:4425385 | B300:11235170H200:3279683 | B300:9493286H200:2027099 |
| Concurrency | B300:~16H200:~13 | B300:~16H200:~8 | B300:~14H200:~5 |
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.
