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 β same interactions as the main inference chart.
Throughput at 19 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T: MI325X hits 479 tok/s/chip, MI355X hits 2200. Per-million costs land at $0.64 and $0.19 respectively. MI355X is 237% cheaper per token; MI355X delivers 359% more tok/s/chip.
MI325X / MI355X on Kimi K2.5/K2.6/K2.7-Code 1T at 28 tok/s/user: 413 / 1748 tok/s/chip, $0.74 / $0.24 per million tokens. MI355X is 211% cheaper per token; MI355X delivers 324% more tok/s/chip.
Toward the upper edge of the 10β45 tok/s/user interactivity band, at 37 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T: MI325X runs 278 tok/s/chip at $1.10/M tokens, MI355X runs 1431 at $0.29/M. MI355X is 277% cheaper per token; MI355X delivers 414% more tok/s/chip. (Numbers reflect the default 8k/1k Β· int4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | MI325X:479.1MI355X:2199.7 | MI325X:412.6MI355X:1747.8 | MI325X:278.2MI355X:1430.8 |
| Cost ($/M tok) | MI325X:$0.638MI355X:$0.189 | MI325X:$0.741MI355X:$0.238 | MI325X:$1.098MI355X:$0.291 |
| tok/s/MW | MI325X:283510MI355X:1052503 | MI325X:244132MI355X:836270 | MI325X:164633MI355X:684609 |
| Concurrency | MI325X:~23MI355X:~52 | MI325X:~14MI355X:~29 | MI325X:~7MI355X:~18 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
TCO $/chip/hr:
Source: SemiAnalysis Market July 2026 Pricing Surveys & AI Cloud TCO Model
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate cost per million tokens per decode chip or per prefill chip, rather than per total chip count. This makes direct token cost comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate input throughput per decode chip or per prefill chip, rather than per total chip count. This makes direct input throughput comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate output throughput per decode chip or per prefill chip, rather than per total chip count. This makes direct output throughput comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate power per decode chip or per prefill chip, rather than per total chip count. This makes direct power comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate Joules per decode chip or per prefill chip, rather than per total chip count. This makes direct Joules per token comparison with aggregated configs not an apples-to-apples comparison.
Matching measurements exist, but their chip series are hidden.
Shift+Scroll to zoom β’ Drag to pan β’ Double-click to reset β’ Click a point to pin tooltip
1 The ATOM engine is promising, however it has yet to serve production tokens. It is still in its infant stage.
Kimi K2.5/2.6/2.7-Code 1T 8K / 1K Total Tokens per $1 TCO vs. Interactivity