MiniMax M3 428B — B200 vs MI325X
Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and MI325X (AMD CDNA 3) on MiniMax M3 428B. 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 52 tok/s/user interactivity on MiniMax M3 428B, B200 delivers 1084 tok/s/GPU at $0.50 per million tokens; MI325X delivers 557 tok/s/GPU at $0.64. B200 is 29% cheaper per token; B200 delivers 95% more tok/s/GPU at this point.
B200 posts 377 tok/s/GPU for $1.44 per million tokens at 95 tok/s/user on MiniMax M3 428B; MI325X posts 209 tok/s/GPU for $1.70. B200 is 18% cheaper per token; B200 delivers 81% more tok/s/GPU.
Throughput at 139 tok/s/user on MiniMax M3 428B: B200 hits 127 tok/s/GPU, MI325X hits 115. Per-million costs land at $4.28 and $3.09 respectively. MI325X is 38% cheaper per token; B200 delivers 11% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | B200:1084.5MI325X:556.9 | B200:377.4MI325X:208.6 | B200:127.0MI325X:114.8 |
| Cost ($/M tok) | B200:$0.496MI325X:$0.640 | B200:$1.438MI325X:$1.697 | B200:$4.278MI325X:$3.092 |
| tok/s/MW | B200:499756MI325X:255452 | B200:173895MI325X:95683 | B200:58533MI325X:52638 |
| Concurrency | B200:~53MI325X:~47 | B200:~8MI325X:~9 | B200:~4MI325X:~4 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.