MiniMax M3 428B — GB200 NVL72 vs GB300 NVL72 Performance per Dollar
Cost per million tokens of GB200 NVL72 (NVIDIA Blackwell) versus GB300 NVL72 (NVIDIA Blackwell) 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.
Push MiniMax M3 428B to 66 tok/s/user and GB200 NVL72 lands at $0.13 per million tokens against GB300 NVL72's $0.04 — GB300 NVL72 pulls ahead by 215%.
GB200 NVL72: $0.27 per million tokens. GB300 NVL72: $0.06. Both at 106 tok/s/user on MiniMax M3 428B, with GB300 NVL72 322% cheaper.
Toward the upper edge of the 27–186 tok/s/user interactivity band — at 147 tok/s/user — GB200 NVL72 runs $0.64 per million tokens on MiniMax M3 428B while GB300 NVL72 runs $0.19. GB300 NVL72 is the cheaper choice by 234%. (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.)
Chip pricing (owning hyperscaler): GB200 NVL72 $1.86/chip/hr · GB300 NVL72 $2.31/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 | GB200 NVL72:$0.128GB300 NVL72:$0.041 | GB200 NVL72:$0.270GB300 NVL72:$0.064 | GB200 NVL72:$0.637GB300 NVL72:$0.191 |
| Concurrency | GB200 NVL72:~222GB300 NVL72:~2048 | GB200 NVL72:~45GB300 NVL72:~368 | GB200 NVL72:~13GB300 NVL72:~34 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.