Kimi K2.5/K2.6/K2.7-Code 1T — H200 vs MI325X Performance per Dollar
Cost per million tokens of H200 (NVIDIA Hopper) versus MI325X (AMD CDNA 3) on Kimi K2.5/K2.6/K2.7-Code 1T. 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 34–51 tok/s/user interactivity band — at 38 tok/s/user — H200 runs $0.77 per million tokens on Kimi K2.5/K2.6/K2.7-Code 1T while MI325X runs $3.08. H200 is the cheaper choice by 299%.
On Kimi K2.5/K2.6/K2.7-Code 1T at 42 tok/s/user, the per-million math comes out to $0.87 for H200 and $3.86 for MI325X; H200 delivers 343% more output per dollar.
At 47 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T, H200 costs $0.99 per million tokens; MI325X costs $4.99. H200 is 405% more cost-efficient at this operating point. (Numbers reflect the default 1k/1k · int4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
Chip pricing (owning hyperscaler): H200 $1.22/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 | H200:$0.771MI325X:$3.077 | H200:$0.870MI325X:$3.859 | H200:$0.988MI325X:$4.994 |
| Concurrency | H200:~50MI325X:~11 | H200:~40MI325X:~8 | H200:~30MI325X:~5 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.