DeepSeek V4 Pro 1.6T — B300 vs H200
Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and H200 (NVIDIA Hopper) 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.
B300 hits 2048 tok/s/GPU for $0.32 per million tokens at 46 tok/s/user on DeepSeek V4 Pro 1.6T. No H200 data at this operating point.
B300: 1095 tok/s/GPU, $0.59 per million tokens at 90 tok/s/user on DeepSeek V4 Pro 1.6T. H200 is unmeasured here.
At 134 tok/s/user on DeepSeek V4 Pro 1.6T, B300 delivers 566 tok/s/GPU at $1.14 per million tokens; H200 hasn't been benchmarked at this target. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | B300:2047.8H200:— | B300:1094.9H200:— | B300:566.1H200:— |
| Cost ($/M tok) | B300:$0.317H200:— | B300:$0.593H200:— | B300:$1.138H200:— |
| tok/s/MW | B300:943678H200:— | B300:504575H200:— | B300:260862H200:— |
| Concurrency | B300:~25H200:— | B300:~6H200:— | B300:~2H200:— |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.