MiniMax M3 428B — B200 vs MI300X Performance per Dollar
Cost per million tokens of B200 (NVIDIA Blackwell) versus MI300X (AMD CDNA 3) on MiniMax M3 428B. 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 13–184 tok/s/user interactivity band — at 55 tok/s/user — B200 runs $0.55 per million tokens on MiniMax M3 428B while MI300X runs $0.87. B200 is the cheaper choice by 59%.
On MiniMax M3 428B at 98 tok/s/user, the per-million math comes out to $1.49 for B200 and $1.86 for MI300X; B200 delivers 25% more output per dollar.
At 142 tok/s/user on MiniMax M3 428B, B200 costs $4.30 per million tokens; MI300X costs $2.87. MI300X is 50% more cost-efficient at this operating point. (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): 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:$0.548MI300X:$0.871 | B200:$1.489MI300X:$1.861 | B200:$4.304MI300X:$2.868 |
| Concurrency | B200:~43MI300X:~29 | B200:~8MI300X:~7 | B200:~4MI300X:~3 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.