DeepSeek R1 — GB200 NVL72 vs H200 Performance per Dollar
Cost per million tokens of GB200 NVL72 (NVIDIA Blackwell) versus H200 (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.
On DeepSeek R1 at 59 tok/s/user, the per-million math comes out to $0.12 for GB200 NVL72 and $0.93 for H200; GB200 NVL72 delivers 655% more output per dollar.
At 98 tok/s/user on DeepSeek R1, GB200 NVL72 costs $0.25 per million tokens; H200 costs $2.66. GB200 NVL72 is 963% more cost-efficient at this operating point.
GB200 NVL72 edges H200 at 136 tok/s/user on DeepSeek R1 — $1.88 per million tokens versus $9.94, a 430% cost-per-token gap. (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.)
GPU pricing (owning hyperscaler): GB200 NVL72 $2.21/GPU/hr · H200 $1.41/GPU/hr. Source: SemiAnalysis Market August 2025 Pricing Surveys & AI Cloud TCO Model.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Dollar per Million Tokens | GB200 NVL72:$0.123H200:$0.929 | GB200 NVL72:$0.250H200:$2.663 | GB200 NVL72:$1.876H200:$9.940 |
| Concurrency | GB200 NVL72:~1335H200:~133 | GB200 NVL72:~627H200:~6 | GB200 NVL72:~108H200:~14 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.