DeepSeek R1 — B300 vs H100 Performance per Dollar
Cost per million tokens of B300 (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.
B300: $0.15 per million tokens. H100: $1.19. Both at 38 tok/s/user on DeepSeek R1, with B300 671% cheaper.
Around the middle of the 9–128 tok/s/user interactivity band — at 68 tok/s/user — B300 runs $0.52 per million tokens on DeepSeek R1 while H100 runs $3.76. B300 is the cheaper choice by 620%.
On DeepSeek R1 at 99 tok/s/user, the per-million math comes out to $1.32 for B300 and $14.1 for H100; B300 delivers 966% more output per dollar. (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): B300 $2.26/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 | B300:$0.154H100:$1.188 | B300:$0.523H100:$3.764 | B300:$1.320H100:$14.069 |
| Concurrency | B300:~2684H100:~640 | B300:~457H100:~68 | B300:~103H100:~8 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.