MiniMax M2.5/M2.7 — GB300 NVL72 vs H200 Performance per Dollar
Cost per million tokens of GB300 NVL72 (NVIDIA Blackwell) versus H200 (NVIDIA Hopper) on MiniMax M2.5/M2.7. 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 M2.5/M2.7 at 49 tok/s/user, the per-million math comes out to $0.13 for GB300 NVL72 and $0.29 for H200; GB300 NVL72 delivers 123% more output per dollar.
At 74 tok/s/user on MiniMax M2.5/M2.7, GB300 NVL72 costs $0.31 per million tokens; H200 costs $0.52. GB300 NVL72 is 68% more cost-efficient at this operating point.
GB300 NVL72 edges H200 at 98 tok/s/user on MiniMax M2.5/M2.7 — $0.74 per million tokens versus $0.91, a 23% 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): GB300 NVL72 $2.31/chip/hr · H200 $1.22/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 | GB300 NVL72:$0.129H200:$0.289 | GB300 NVL72:$0.308H200:$0.518 | GB300 NVL72:$0.736H200:$0.906 |
| Concurrency | GB300 NVL72:~512H200:~51 | GB300 NVL72:~128H200:~18 | GB300 NVL72:~32H200:~8 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.