MiniMax M3 428B · GPU comparison

MiniMax M3 428B — MI300X vs MI355X

Head-to-head AI inference benchmark comparison of MI300X (AMD CDNA 3) and MI355X (AMD CDNA 4) 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 44 tok/s/user interactivity on MiniMax M3 428B, MI300X delivers 738 tok/s/GPU at $0.42 per million tokens; MI355X delivers 6306 tok/s/GPU at $0.07. MI355X is 544% cheaper per token; MI355X delivers 755% more tok/s/GPU at this point.

MI300X posts 594 tok/s/GPU for $0.52 per million tokens at 83 tok/s/user on MiniMax M3 428B; MI355X posts 4221 tok/s/GPU for $0.10. MI355X is 440% cheaper per token; MI355X delivers 610% more tok/s/GPU.

Throughput at 123 tok/s/user on MiniMax M3 428B: MI300X hits 352 tok/s/GPU, MI355X hits 3108. Per-million costs land at $0.86 and $0.13 respectively. MI355X is 550% cheaper per token; MI355X delivers 783% more tok/s/GPU. (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.)

View performance-per-dollar view →

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/gpu)
MI300X:737.7MI355X:6305.6
MI300X:594.2MI355X:4221.1
MI300X:352.0MI355X:3108.4
Cost ($/M tok)
MI300X:$0.422MI355X:$0.065
MI300X:$0.523MI355X:$0.097
MI300X:$0.864MI355X:$0.133
tok/s/MW
MI300X:530740MI355X:3017020
MI300X:427449MI355X:2019684
MI300X:253227MI355X:1487295
Concurrency
MI300X:~16MI355X:~109
MI300X:~8MI355X:~27
MI300X:~3MI355X:~13

Inference Performance

Inference performance metrics across different models, hardware configurations, and serving parameters.

Vendor:
Aggregation:
Spec Decoding: