DeepSeek V4 Pro 1.6T — B200 vs GB300 NVL72
Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and GB300 NVL72 (NVIDIA Blackwell) on DeepSeek V4 Pro 1.6T. 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 68 tok/s/user on DeepSeek V4 Pro 1.6T: B200 hits 3177 tok/s/GPU, GB300 NVL72 hits 9759. Per-million costs land at $0.17 and $0.08 respectively. GB300 NVL72 is 124% cheaper per token; GB300 NVL72 delivers 207% more tok/s/GPU.
B200 / GB300 NVL72 on DeepSeek V4 Pro 1.6T at 121 tok/s/user: 840 / 3816 tok/s/GPU, $0.64 / $0.20 per million tokens. GB300 NVL72 is 227% cheaper per token; GB300 NVL72 delivers 355% more tok/s/GPU.
Toward the upper edge of the 15–226 tok/s/user interactivity band, at 173 tok/s/user on DeepSeek V4 Pro 1.6T: B200 runs 470 tok/s/GPU at $1.18/M tokens, GB300 NVL72 runs 895 at $0.84/M. GB300 NVL72 is 40% cheaper per token; GB300 NVL72 delivers 91% more tok/s/GPU. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | B200:3177.3GB300 NVL72:9758.8 | B200:839.6GB300 NVL72:3816.3 | B200:469.5GB300 NVL72:894.8 |
| Cost ($/M tok) | B200:$0.169GB300 NVL72:$0.075 | B200:$0.640GB300 NVL72:$0.196 | B200:$1.180GB300 NVL72:$0.841 |
| tok/s/MW | B200:1464199GB300 NVL72:4647034 | B200:386901GB300 NVL72:1817290 | B200:216382GB300 NVL72:426078 |
| Concurrency | B200:~56GB300 NVL72:~1024 | B200:~7GB300 NVL72:~315 | B200:~3GB300 NVL72:~29 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.