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 52 tok/s/user and GB200 NVL72 lands at $0.32 per million tokens against GB300 NVL72's $0.29 — GB300 NVL72 pulls ahead by 10%.
GB200 NVL72: $0.81 per million tokens. GB300 NVL72: $0.96. Both at 86 tok/s/user on MiniMax M3 428B, with GB200 NVL72 18% cheaper.
Toward the upper edge of the 17–156 tok/s/user interactivity band — at 121 tok/s/user — GB200 NVL72 runs $2.37 per million tokens on MiniMax M3 428B while GB300 NVL72 runs $2.43. GB200 NVL72 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): 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:$0.322GB300 NVL72:$0.293 | GB200 NVL72:$0.815GB300 NVL72:$0.960 | GB200 NVL72:$2.372GB300 NVL72:$2.432 |
| Concurrency | GB200 NVL72:~180GB300 NVL72:~452 | GB200 NVL72:~76GB300 NVL72:~72 | GB200 NVL72:~19GB300 NVL72:~21 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.