Kimi K2.5/K2.6/K2.7-Code 1T — B300 vs GB300 NVL72
Head-to-head AI inference benchmark comparison of B300 (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.
B300 / GB300 NVL72 on Kimi K2.5/K2.6/K2.7-Code 1T at 64 tok/s/user: 941 / 1678 tok/s/GPU, $0.69 / $0.57 per million tokens. GB300 NVL72 is 22% cheaper per token; GB300 NVL72 delivers 78% more tok/s/GPU.
Around the middle of the 35–153 tok/s/user interactivity band, at 94 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T: B300 runs 910 tok/s/GPU at $0.72/M tokens, GB300 NVL72 runs 399 at $1.83/M. B300 is 156% cheaper per token; B300 delivers 128% more tok/s/GPU.
Setting 124 tok/s/user as the target on Kimi K2.5/K2.6/K2.7-Code 1T, B300 produces 759 tok/s/GPU ($0.85 per million tokens) and GB300 NVL72 produces 208 ($3.54). B300 is 315% cheaper per token; B300 delivers 265% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | B300:940.6GB300 NVL72:1677.6 | B300:910.0GB300 NVL72:398.9 | B300:759.3GB300 NVL72:208.0 |
| Cost ($/M tok) | B300:$0.691GB300 NVL72:$0.566 | B300:$0.715GB300 NVL72:$1.830 | B300:$0.854GB300 NVL72:$3.540 |
| tok/s/MW | B300:433436GB300 NVL72:798844 | B300:419364GB300 NVL72:189965 | B300:349903GB300 NVL72:99036 |
| Concurrency | B300:~30GB300 NVL72:~464 | B300:~21GB300 NVL72:~73 | B300:~12GB300 NVL72:~29 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.