DeepSeek R1 — B200 vs B300 Performance per Dollar
Cost per million tokens of B200 (NVIDIA Blackwell) versus B300 (NVIDIA Blackwell) 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.
Push DeepSeek R1 to 78 tok/s/user and B200 lands at $0.58 per million tokens against B300's $0.96 — B200 pulls ahead by 65%.
B200: $1.91 per million tokens. B300: $2.29. Both at 143 tok/s/user on DeepSeek R1, with B200 20% cheaper.
Toward the upper edge of the 14–272 tok/s/user interactivity band — at 207 tok/s/user — B200 runs $5.14 per million tokens on DeepSeek R1 while B300 runs $5.07. B300 is the cheaper choice by 2%. (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 · B300 $2.26/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:$0.583B300:$0.961 | B200:$1.914B300:$2.289 | B200:$5.143B300:$5.067 |
| Concurrency | B200:~42B300:~61 | B200:~41B300:~8 | B200:~2B300:~3 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.