MiniMax M3 428B — H200 vs MI300X
Head-to-head AI inference benchmark comparison of H200 (NVIDIA Hopper) and MI300X (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.
Setting 55 tok/s/user as the target on MiniMax M3 428B, H200 produces 932 tok/s/GPU ($0.42 per million tokens) and MI300X produces 362 ($0.87). H200 is 106% cheaper per token; H200 delivers 157% more tok/s/GPU.
At 98 tok/s/user interactivity on MiniMax M3 428B, H200 delivers 553 tok/s/GPU at $0.72 per million tokens; MI300X delivers 167 tok/s/GPU at $1.86. H200 is 160% cheaper per token; H200 delivers 232% more tok/s/GPU at this point.
H200 posts 351 tok/s/GPU for $1.09 per million tokens at 142 tok/s/user on MiniMax M3 428B; MI300X posts 109 tok/s/GPU for $2.87. H200 is 164% cheaper per token; H200 delivers 222% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | H200:932.0MI300X:361.9 | H200:553.4MI300X:166.8 | H200:351.2MI300X:109.0 |
| Cost ($/M tok) | H200:$0.424MI300X:$0.871 | H200:$0.716MI300X:$1.861 | H200:$1.088MI300X:$2.868 |
| tok/s/MW | H200:538716MI300X:202206 | H200:319862MI300X:93205 | H200:203022MI300X:60907 |
| Concurrency | H200:~36MI300X:~29 | H200:~12MI300X:~7 | H200:~5MI300X:~3 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.