Head-to-head AI inference benchmark comparison of MI300X (AMD CDNA 3) and MI325X (AMD CDNA 3) on DeepSeek R1. 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.
MI300X posts 710 tok/s/chip for $0.37 per million tokens at 24 tok/s/user on DeepSeek R1; MI325X posts 881 tok/s/chip for $0.35. MI325X is 7% cheaper per token; MI325X delivers 24% more tok/s/chip.
Throughput at 33 tok/s/user on DeepSeek R1: MI300X hits 586 tok/s/chip, MI325X hits 724. Per-million costs land at $0.45 and $0.42 respectively. MI325X is 7% cheaper per token; MI325X delivers 24% more tok/s/chip.
MI300X / MI325X on DeepSeek R1 at 42 tok/s/user: 406 / 526 tok/s/chip, $0.65 / $0.58 per million tokens. MI325X is 12% cheaper per token; MI325X delivers 30% more tok/s/chip. (Numbers reflect the default 8k/1k Β· fp8 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | MI300X:710.4MI325X:881.0 | MI300X:585.6MI325X:724.5 | MI300X:406.3MI325X:526.3 |
| Cost ($/M tok) | MI300X:$0.371MI325X:$0.347 | MI300X:$0.451MI325X:$0.422 | MI300X:$0.650MI325X:$0.581 |
| tok/s/MW | MI300X:511113MI325X:521287 | MI300X:421292MI325X:428669 | MI300X:292288MI325X:311408 |
| Concurrency | MI300X:~28MI325X:~36 | MI300X:~18MI325X:~21 | MI300X:~10MI325X:~12 |
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.
No measurements to plot for this selection. Review the benchmark controls above or adjust quick filters.
Shift+Scroll to zoom β’ Drag to pan β’ Double-click to reset β’ Click a point to pin tooltip
DeepSeek R1 0528 671B 8K / 1K Total Tokens per $1 TCO vs. Interactivity