Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and H100 (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 →
B200 / H100 on DeepSeek V4.1 Flash 552B at 82 tok/s/user: 70537 / 10876 tok/s/chip, $0.01 / $0.03 per million tokens. B200 is 339% cheaper per token; B200 delivers 549% more tok/s/chip.
Around the middle of the 8–303 tok/s/user interactivity band, at 156 tok/s/user on DeepSeek V4.1 Flash 552B: B200 runs 51613 tok/s/chip at $0.01/M tokens, H100 runs 6876 at $0.05/M. B200 is 408% cheaper per token; B200 delivers 651% more tok/s/chip.
Setting 230 tok/s/user as the target on DeepSeek V4.1 Flash 552B, B200 produces 30786 tok/s/chip ($0.02 per million tokens) and H100 produces 4187 ($0.08). B200 is 397% cheaper per token; B200 delivers 635% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B200:70536.7H100:10876.0 | B200:51612.9H100:6876.0 | B200:30786.3H100:4187.3 |
| Cost ($/M tok) | B200:$0.007H100:$0.030 | B200:$0.009H100:$0.047 | B200:$0.016H100:$0.078 |
| tok/s/MW | B200:41249556H100:7938712 | B200:30182969H100:5018983 | B200:18003673H100:3056440 |
| Concurrency | B200:~64H100:~15 | B200:~14H100:~8 | B200:~8H100:~4 |
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.
