DeepSeek R1 — B200 vs MI300X Performance per Dollar
Cost per million tokens of B200 (NVIDIA Blackwell) versus MI300X (AMD CDNA 3) 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.
Near the low end of the 14–51 tok/s/user interactivity band — at 23 tok/s/user — B200 runs $0.09 per million tokens on DeepSeek R1 while MI300X runs $0.43. B200 is the cheaper choice by 364%.
On DeepSeek R1 at 32 tok/s/user, the per-million math comes out to $0.10 for B200 and $0.51 for MI300X; B200 delivers 415% more output per dollar.
At 42 tok/s/user on DeepSeek R1, B200 costs $0.10 per million tokens; MI300X costs $0.75. B200 is 615% more cost-efficient at this operating point. (Numbers reflect the default 8k/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): B200 $1.95/GPU/hr · MI300X $1.12/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 | B200:$0.092MI300X:$0.429 | B200:$0.099MI300X:$0.510 | B200:$0.105MI300X:$0.750 |
| Concurrency | B200:~1088MI300X:~30 | B200:~1005MI300X:~18 | B200:~557MI300X:~10 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.