DeepSeek R1 · Chip comparison
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 35 tok/s/user as the target on DeepSeek R1, MI325X produces 966 tok/s/chip ($0.32 per million tokens) and MI355X produces 2,657 ($0.16). MI355X is 50% cheaper per token; MI355X delivers 175% more tok/s/chip.
At 57 tok/s/user interactivity on DeepSeek R1, MI325X delivers 747 tok/s/chip at $0.41 per million tokens; MI355X delivers 2,050 tok/s/chip at $0.20. MI355X is 50% cheaper per token; MI355X delivers 174% more tok/s/chip at this point.
MI325X posts 607 tok/s/chip for $0.50 per million tokens at 78 tok/s/user on DeepSeek R1; MI355X posts 1,529 tok/s/chip for $0.27. MI355X is 46% cheaper per token; MI355X delivers 152% more tok/s/chip. (Numbers reflect the default 8k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | MI325X966MI355X2,657 | MI325X747MI355X2,050 | MI325X607MI355X1,529 |
| Cost ($/M tok) | MI325X$0.32MI355X$0.16 | MI325X$0.41MI355X$0.20 | MI325X$0.50MI355X$0.27 |
| tok/s/MW | MI325X571,610MI355X1,271,122 | MI325X442,003MI355X980,633 | MI325X358,995MI355X731,357 |
| Concurrency | MI325X~24MI355X~120 | MI325X~11MI355X~32 | MI325X~7MI355X~34 |