MiniMax M2.5/M2.7 — GB200 NVL72 vs H200
Head-to-head AI inference benchmark comparison of GB200 NVL72 (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.
GB200 NVL72 posts 3623 tok/s/chip for $0.14 per million tokens at 52 tok/s/user on MiniMax M2.5/M2.7; H200 posts 1095 tok/s/chip for $0.32. GB200 NVL72 is 123% cheaper per token; GB200 NVL72 delivers 231% more tok/s/chip.
Throughput at 80 tok/s/user on MiniMax M2.5/M2.7: GB200 NVL72 hits 1276 tok/s/chip, H200 hits 541. Per-million costs land at $0.42 and $0.58 respectively. GB200 NVL72 is 39% cheaper per token; GB200 NVL72 delivers 136% more tok/s/chip.
GB200 NVL72 / H200 on MiniMax M2.5/M2.7 at 107 tok/s/user: 667 / 303 tok/s/chip, $0.78 / $1.11 per million tokens. GB200 NVL72 is 42% cheaper per token; GB200 NVL72 delivers 120% more tok/s/chip. (Numbers reflect the default 1k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL72:3622.7H200:1095.1 | GB200 NVL72:1276.4H200:540.6 | GB200 NVL72:667.3H200:303.3 |
| Cost ($/M tok) | GB200 NVL72:$0.142H200:$0.317 | GB200 NVL72:$0.419H200:$0.582 | GB200 NVL72:$0.778H200:$1.108 |
| tok/s/MW | GB200 NVL72:1937257H200:799362 | GB200 NVL72:682586H200:394579 | GB200 NVL72:356853H200:221394 |
| Concurrency | GB200 NVL72:~489H200:~42 | GB200 NVL72:~64H200:~15 | GB200 NVL72:~32H200:~6 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.