Kimi K2.5/K2.6/K2.7-Code 1T — B200 vs GB300 NVL72
Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and GB300 NVL72 (NVIDIA Blackwell) on Kimi K2.5/K2.6/K2.7-Code 1T. 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 64 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T: B200 hits 857 tok/s/GPU, GB300 NVL72 hits 1678. Per-million costs land at $0.63 and $0.57 respectively. GB300 NVL72 is 11% cheaper per token; GB300 NVL72 delivers 96% more tok/s/GPU.
B200 / GB300 NVL72 on Kimi K2.5/K2.6/K2.7-Code 1T at 94 tok/s/user: 480 / 399 tok/s/GPU, $1.12 / $1.83 per million tokens. B200 is 64% cheaper per token; B200 delivers 20% more tok/s/GPU.
Toward the upper edge of the 35–153 tok/s/user interactivity band, at 124 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T: B200 runs 263 tok/s/GPU at $2.03/M tokens, GB300 NVL72 runs 208 at $3.54/M. B200 is 74% cheaper per token; B200 delivers 27% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | B200:856.5GB300 NVL72:1677.6 | B200:480.2GB300 NVL72:398.9 | B200:263.4GB300 NVL72:208.0 |
| Cost ($/M tok) | B200:$0.629GB300 NVL72:$0.566 | B200:$1.117GB300 NVL72:$1.830 | B200:$2.033GB300 NVL72:$3.540 |
| tok/s/MW | B200:394722GB300 NVL72:798844 | B200:221278GB300 NVL72:189965 | B200:121399GB300 NVL72:99036 |
| Concurrency | B200:~29GB300 NVL72:~464 | B200:~10GB300 NVL72:~73 | B200:~4GB300 NVL72:~29 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.