MiniMax M2.5/M2.7 · Chip comparison

MiniMax M2.5/M2.7 — H200 vs MI355X

Head-to-head AI inference benchmark comparison of H200 (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.

Throughput at 36 tok/s/user on MiniMax M2.5/M2.7: H200 hits 3,221 tok/s/chip, MI355X hits 5,471. Costs per million tokens land at $0.11 and $0.076 respectively. MI355X is 28% cheaper per token; MI355X delivers 70% more tok/s/chip.

H200 / MI355X on MiniMax M2.5/M2.7 at 61 tok/s/user: 2,197 / 3,197 tok/s/chip, $0.15 / $0.13 per million tokens. MI355X is 16% cheaper per token; MI355X delivers 46% more tok/s/chip.

Toward the upper edge of the 11–110 tok/s/user interactivity band, at 86 tok/s/user on MiniMax M2.5/M2.7: H200 runs 1,452 tok/s/chip at $0.23 per million tokens, MI355X runs 1,954 at $0.21. MI355X is 9% cheaper per token; MI355X delivers 35% more tok/s/chip. (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 →

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/chip)
H2003,221MI355X5,471
H2002,197MI355X3,197
H2001,452MI355X1,954
Cost ($/M tok)
H200$0.11MI355X$0.076
H200$0.15MI355X$0.13
H200$0.23MI355X$0.21
tok/s/MW
H2002,350,935MI355X2,617,800
H2001,603,619MI355X1,529,799
H2001,059,771MI355X934,901
Concurrency
H200~40MI355X~72
H200~17MI355X~12
H200~8MI355X~5