MiniMax M2.5/M2.7 · GPU comparison

MiniMax M2.5/M2.7 — H100 vs MI355X

Head-to-head AI inference benchmark comparison of H100 (NVIDIA Hopper) and MI355X (AMD CDNA 4) 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.

H100 posts 614 tok/s/GPU for $0.58 per million tokens at 59 tok/s/user on MiniMax M2.5/M2.7; MI355X posts 1619 tok/s/GPU for $0.26. MI355X is 119% cheaper per token; MI355X delivers 164% more tok/s/GPU.

Throughput at 78 tok/s/user on MiniMax M2.5/M2.7: H100 hits 378 tok/s/GPU, MI355X hits 883. Per-million costs land at $0.94 and $0.47 respectively. MI355X is 100% cheaper per token; MI355X delivers 133% more tok/s/GPU.

H100 / MI355X on MiniMax M2.5/M2.7 at 98 tok/s/user: 199 / 503 tok/s/GPU, $1.82 / $0.78 per million tokens. MI355X is 133% cheaper per token; MI355X delivers 153% 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.)

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)
H100:614.1MI355X:1618.9
H100:378.2MI355X:882.7
H100:199.2MI355X:503.3
Cost ($/M tok)
H100:$0.580MI355X:$0.264
H100:$0.937MI355X:$0.469
H100:$1.824MI355X:$0.782
tok/s/MW
H100:354968MI355X:610889
H100:218629MI355X:333085
H100:115163MI355X:189919
Concurrency
H100:~42MI355X:~52
H100:~12MI355X:~22
H100:~8MI355X:~5

Inference Performance

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