MiniMax M2.5/M2.7 — B300 vs H200
Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and H200 (NVIDIA Hopper) 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 60 tok/s/user on MiniMax M2.5/M2.7: B300 hits 3285 tok/s/GPU, H200 hits 913. Per-million costs land at $0.20 and $0.43 respectively. B300 is 118% cheaper per token; B300 delivers 260% more tok/s/GPU.
B300 / H200 on MiniMax M2.5/M2.7 at 82 tok/s/user: 1556 / 402 tok/s/GPU, $0.42 / $0.95 per million tokens. B300 is 123% cheaper per token; B300 delivers 287% more tok/s/GPU.
Toward the upper edge of the 39–125 tok/s/user interactivity band, at 104 tok/s/user on MiniMax M2.5/M2.7: B300 runs 947 tok/s/GPU at $0.66/M tokens, H200 runs 227 at $1.72/M. B300 is 160% cheaper per token; B300 delivers 317% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | B300:3284.7H200:913.4 | B300:1556.2H200:402.4 | B300:946.7H200:226.8 |
| Cost ($/M tok) | B300:$0.197H200:$0.430 | B300:$0.424H200:$0.945 | B300:$0.664H200:$1.725 |
| tok/s/MW | B300:1513669H200:527982 | B300:717138H200:232615 | B300:436265H200:131109 |
| Concurrency | B300:~112H200:~62 | B300:~39H200:~20 | B300:~19H200:~9 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.