DeepSeek R1 — MI325X vs MI355X
Head-to-head AI inference benchmark comparison of MI325X (AMD CDNA 3) and MI355X (AMD CDNA 4) on DeepSeek R1. 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.
Setting 25 tok/s/user as the target on DeepSeek R1, MI325X produces 854 tok/s/chip ($0.36 per million tokens) and MI355X produces 3471 ($0.12). MI355X is 198% cheaper per token; MI355X delivers 306% more tok/s/chip.
At 35 tok/s/user interactivity on DeepSeek R1, MI325X delivers 677 tok/s/chip at $0.45 per million tokens; MI355X delivers 2454 tok/s/chip at $0.17. MI355X is 166% cheaper per token; MI355X delivers 262% more tok/s/chip at this point.
MI325X posts 457 tok/s/chip for $0.67 per million tokens at 44 tok/s/user on DeepSeek R1; MI355X posts 2206 tok/s/chip for $0.19. MI355X is 254% cheaper per token; MI355X delivers 383% more tok/s/chip. (Numbers reflect the default 8k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | MI325X:854.2MI355X:3471.5 | MI325X:677.5MI355X:2454.1 | MI325X:456.8MI355X:2205.6 |
| Cost ($/M tok) | MI325X:$0.358MI355X:$0.120 | MI325X:$0.451MI355X:$0.170 | MI325X:$0.669MI355X:$0.189 |
| tok/s/MW | MI325X:505466MI355X:1660986 | MI325X:400884MI355X:1174226 | MI325X:270289MI355X:1055316 |
| Concurrency | MI325X:~33MI355X:~2048 | MI325X:~19MI355X:~61 | MI325X:~10MI355X:~41 |
Total Tokens per $1 USD (Owning - Neocloud Giant) vs. Interactivity
DeepSeek R1 0528 671B • FP4 • 8K / 1K • Source: SemiAnalysis InferenceX™
DeepSeek R1 0528 671B • FP4 • 8K / 1K • Source: SemiAnalysis InferenceX™TCO $/chip/hr: VR200: 3.61H100: 1.55H200: 1.59B200: 2.07B300: 2.52GB200: 2.26GB300: 2.79MI300X: 1.16MI325X: 1.32MI355X: 2.09RTX6000PRO: 0.75Jalapeño (Teacup): 1.56
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 data available
Please change the model, sequence, precision, date range or chip selection.
Shift+Scroll to zoom • Drag to pan • Double-click to reset • Click a point to pin tooltip
Interactivity is the rate at which a single user receives generated tokens while the model streams its answer — how quickly new words appear on screen. Higher values feel snappier; operators trade it against batch throughput.
How many total tokens (input + output) one US dollar of infrastructure spend buys, priced with the all-in hourly ownership cost of a Neocloud Giant operator. It is the reciprocal of cost per token, so higher means cheaper.
Formula: tok/$ = (total tok/s/chip × 3,600) ÷ all-in cost per chip-hour ($)
1 The ATOM engine is promising, however it has yet to serve production tokens. It is still in its infant stage.