MiniMax M3 428B — GB200 NVL72 vs MI325X Performance per Dollar
Cost per million tokens of GB200 NVL72 (NVIDIA Blackwell) versus MI325X (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.
On MiniMax M3 428B at 52 tok/s/user, the per-million math comes out to $0.32 for GB200 NVL72 and $0.55 for MI325X; GB200 NVL72 delivers 70% more output per dollar.
At 87 tok/s/user on MiniMax M3 428B, GB200 NVL72 costs $0.81 per million tokens; MI325X costs $1.31. GB200 NVL72 is 61% more cost-efficient at this operating point.
MI325X edges GB200 NVL72 at 122 tok/s/user on MiniMax M3 428B — $2.13 per million tokens versus $2.45, a 15% cost-per-token gap. (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 · MI325X $1.10/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:$0.323MI325X:$0.549 | GB200 NVL72:$0.809MI325X:$1.305 | GB200 NVL72:$2.449MI325X:$2.135 |
| Concurrency | GB200 NVL72:~180MI325X:~47 | GB200 NVL72:~74MI325X:~11 | GB200 NVL72:~18MI325X:~5 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.