Kimi K2.5/K2.6 1T — B200 vs MI355X Performance per Dollar
Cost per million tokens of B200 (NVIDIA Blackwell) versus MI355X (AMD CDNA 4) on Kimi K2.5/K2.6 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 35–113 tok/s/user interactivity band — at 54 tok/s/user — B200 runs $0.52 per million tokens on Kimi K2.5/K2.6 1T while MI355X runs $0.66. B200 is the cheaper choice by 26%.
Cost-per-million is essentially even between B200 ($0.79) and MI355X ($0.80) at 74 tok/s/user on Kimi K2.5/K2.6 1T.
At 94 tok/s/user on Kimi K2.5/K2.6 1T, B200 costs $1.08 per million tokens; MI355X costs $1.04. MI355X is 4% more cost-efficient at this operating point. (Numbers reflect the default 1k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
GPU pricing (owning hyperscaler): B200 $1.95/GPU/hr · MI355X $1.48/GPU/hr. Source: SemiAnalysis Market August 2025 Pricing Surveys & AI Cloud TCO Model.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Dollar per Million Tokens | B200:$0.521MI355X:$0.658 | B200:$0.794MI355X:$0.797 | B200:$1.078MI355X:$1.040 |
| Concurrency | B200:~42MI355X:~24 | B200:~19MI355X:~14 | B200:~10MI355X:~9 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.