Head-to-head AI inference benchmark comparison of B200 (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 →
Near the low end of the 8–320 tok/s/user interactivity band, at 86 tok/s/user on DeepSeek V4.1 Flash 552B: B200 runs 67556 tok/s/chip at $0.01/M tokens, H200 runs 18842 at $0.02/M. B200 is 153% cheaper per token; B200 delivers 259% more tok/s/chip.
Setting 164 tok/s/user as the target on DeepSeek V4.1 Flash 552B, B200 produces 49478 tok/s/chip ($0.01 per million tokens) and H200 produces 10457 ($0.03). B200 is 234% cheaper per token; B200 delivers 373% more tok/s/chip.
At 242 tok/s/user interactivity on DeepSeek V4.1 Flash 552B, B200 delivers 28432 tok/s/chip at $0.02 per million tokens; H200 delivers 5431 tok/s/chip at $0.06. B200 is 269% cheaper per token; B200 delivers 424% more tok/s/chip at this point. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B200:67556.1H200:18841.6 | B200:49478.4H200:10457.2 | B200:28432.3H200:5430.8 |
| Cost ($/M tok) | B200:$0.007H200:$0.018 | B200:$0.010H200:$0.032 | B200:$0.017H200:$0.062 |
| tok/s/MW | B200:39506470H200:13752981 | B200:28934723H200:7632978 | B200:16627067H200:3964073 |
| Concurrency | B200:~64H200:~13 | B200:~13H200:~6 | B200:~7H200:~2 |
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.
