DeepSeek R1 — B200 vs H100 Performance per Dollar
Cost per million tokens of B200 (NVIDIA Blackwell) versus H100 (NVIDIA Hopper) on DeepSeek R1. Owning-hyperscaler TCO normalized by output tokens — performance per dollar across LLM workloads. Pick the more cost-efficient SKU at every target interactivity level. Use the chart controls below to switch sequences, precisions, and metrics — same interactions as the main inference chart.
At 42 tok/s/user on DeepSeek R1, B200 costs $0.10 per million tokens; H100 costs $1.23. B200 is 1127% more cost-efficient at this operating point.
B200 edges H100 at 71 tok/s/user on DeepSeek R1 — $0.48 per million tokens versus $4.38, a 807% cost-per-token gap.
Push DeepSeek R1 to 100 tok/s/user and B200 lands at $0.97 per million tokens against H100's $14.1 — B200 pulls ahead by 1363%. (Numbers reflect the default 1k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
Chip pricing (owning hyperscaler): B200 $1.73/chip/hr · H100 $1.17/chip/hr. Source: SemiAnalysis Market July 2026 Pricing Surveys & AI Cloud TCO Model.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Dollar per Million Tokens | B200:$0.100H100:$1.228 | B200:$0.482H100:$4.378 | B200:$0.965H100:$14.123 |
| Concurrency | B200:~1577H100:~558 | B200:~59H100:~52 | B200:~139H100:~8 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.