Kimi K2.5/K2.6/K2.7-Code 1T — MI325X vs MI355X
Head-to-head AI inference benchmark comparison of MI325X (AMD CDNA 3) and MI355X (AMD CDNA 4) on Kimi K2.5/K2.6/K2.7-Code 1T. Latency, throughput, and cost across LLM workloads. Use the chart controls below to switch sequences, precisions, and metrics — same interactions as the main inference chart.
Throughput at 27 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T: MI325X hits 140 tok/s/GPU, MI355X hits 825. Per-million costs land at $2.55 and $0.50 respectively. MI355X is 413% cheaper per token; MI355X delivers 488% more tok/s/GPU.
MI325X / MI355X on Kimi K2.5/K2.6/K2.7-Code 1T at 35 tok/s/user: 118 / 608 tok/s/GPU, $3.02 / $0.68 per million tokens. MI355X is 346% cheaper per token; MI355X delivers 417% more tok/s/GPU.
Toward the upper edge of the 20–51 tok/s/user interactivity band, at 43 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T: MI325X runs 75 tok/s/GPU at $4.73/M tokens, MI355X runs 466 at $0.88/M. MI355X is 438% cheaper per token; MI355X delivers 520% more tok/s/GPU. (Numbers reflect the default 1k/1k · int4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | MI325X:140.4MI355X:825.2 | MI325X:117.6MI355X:607.5 | MI325X:75.1MI355X:465.6 |
| Cost ($/M tok) | MI325X:$2.553MI355X:$0.497 | MI325X:$3.024MI355X:$0.678 | MI325X:$4.731MI355X:$0.879 |
| tok/s/MW | MI325X:64411MI355X:311393 | MI325X:53958MI355X:229263 | MI325X:34469MI355X:175683 |
| Concurrency | MI325X:~21MI355X:~62 | MI325X:~14MI355X:~36 | MI325X:~7MI355X:~22 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.