MiniMax M2.5/M2.7 — MI300X vs MI325X
Head-to-head AI inference benchmark comparison of MI300X (AMD CDNA 3) and MI325X (AMD CDNA 3) on MiniMax M2.5/M2.7. 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.
At 44 tok/s/user interactivity on MiniMax M2.5/M2.7, MI300X delivers 1249 tok/s/GPU at $0.25 per million tokens; MI325X delivers 1782 tok/s/GPU at $0.20. MI325X is 25% cheaper per token; MI325X delivers 43% more tok/s/GPU at this point.
MI300X posts 779 tok/s/GPU for $0.40 per million tokens at 61 tok/s/user on MiniMax M2.5/M2.7; MI325X posts 910 tok/s/GPU for $0.39. MI325X is 3% cheaper per token; MI325X delivers 17% more tok/s/GPU.
Throughput at 78 tok/s/user on MiniMax M2.5/M2.7: MI300X hits 412 tok/s/GPU, MI325X hits 494. Per-million costs land at $0.75 and $0.74 respectively. MI325X is 1% cheaper per token; MI325X delivers 20% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | MI300X:1248.6MI325X:1782.0 | MI300X:778.8MI325X:910.0 | MI300X:411.9MI325X:494.1 |
| Cost ($/M tok) | MI300X:$0.250MI325X:$0.200 | MI300X:$0.402MI325X:$0.391 | MI300X:$0.747MI325X:$0.739 |
| tok/s/MW | MI300X:697553MI325X:817431 | MI300X:435091MI325X:417438 | MI300X:230105MI325X:226640 |
| Concurrency | MI300X:~63MI325X:~166 | MI300X:~26MI325X:~61 | MI300X:~11MI325X:~27 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.