MiniMax M3 428B — H100 vs MI300X Performance per Dollar
Cost per million tokens of H100 (NVIDIA Hopper) versus MI300X (AMD CDNA 3) on MiniMax M3 428B. Owning-hyperscaler TCO normalized by output tokens — performance per dollar across LLM workloads. Pick the more cost-efficient SKU at every target interactivity level. Use the chart controls below to switch sequences, precisions, and metrics — same interactions as the main inference chart.
H100 edges MI300X at 48 tok/s/user on MiniMax M3 428B — $0.41 per million tokens versus $0.60, a 48% cost-per-token gap.
Push MiniMax M3 428B to 82 tok/s/user and H100 lands at $0.76 per million tokens against MI300X's $1.54 — H100 pulls ahead by 103%.
H100: $1.23 per million tokens. MI300X: $2.14. Both at 115 tok/s/user on MiniMax M3 428B, with H100 74% cheaper. (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.)
GPU pricing (owning hyperscaler): H100 $1.30/GPU/hr · MI300X $1.12/GPU/hr. Source: SemiAnalysis Market August 2025 Pricing Surveys & AI Cloud TCO Model.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Dollar per Million Tokens | H100:$0.406MI300X:$0.600 | H100:$0.758MI300X:$1.541 | H100:$1.234MI300X:$2.145 |
| Concurrency | H100:~83MI300X:~45 | H100:~26MI300X:~10 | H100:~11MI300X:~5 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.