MiniMax M2.5/M2.7 · Chip comparison

MiniMax M2.5/M2.7 — H200 vs MI300X

Head-to-head AI inference benchmark comparison of H200 (NVIDIA Hopper) and MI300X (AMD CDNA 3) 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.

Near the low end of the 11–84 tok/s/user interactivity band, at 29 tok/s/user on MiniMax M2.5/M2.7: H200 runs 3,752 tok/s/chip at $0.090 per million tokens, MI300X runs 1,578 at $0.17. H200 is 46% cheaper per token; H200 delivers 138% more tok/s/chip.

Setting 48 tok/s/user as the target on MiniMax M2.5/M2.7, H200 produces 2,595 tok/s/chip ($0.13 per million tokens) and MI300X produces 1,412 ($0.19). H200 is 30% cheaper per token; H200 delivers 84% more tok/s/chip.

At 66 tok/s/user interactivity on MiniMax M2.5/M2.7, H200 delivers 2,038 tok/s/chip at $0.17 per million tokens; MI300X delivers 1,086 tok/s/chip at $0.24. H200 is 32% cheaper per token; H200 delivers 88% more tok/s/chip at this point. (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)
H2003,752MI300X1,578
H2002,595MI300X1,412
H2002,038MI300X1,086
Cost ($/M tok)
H200$0.090MI300X$0.17
H200$0.13MI300X$0.19
H200$0.17MI300X$0.24
tok/s/MW
H2002,738,410MI300X1,135,484
H2001,893,989MI300X1,015,827
H2001,487,257MI300X781,024
Concurrency
H200~58MI300X~21
H200~25MI300X~7
H200~14MI300X~4