GPU comparison

B300 vs H200

Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and H200 (NVIDIA Hopper). Latency, throughput, and cost across LLM workloads. Use the chart controls below to switch models, sequences, precisions, and metrics — same interactions as the main inference chart.

Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
Metric
Interactivity (tok/s/user)
Interactivity (tok/s/user)
Interactivity (tok/s/user)
Throughput (tok/s/gpu)
B300:5900.6H200:1652.8
B300:2308.2H200:635.1
B300:1191.8H200:292.1
Cost ($/M tok)
B300:$0.111H200:$0.236
B300:$0.275H200:$0.593
B300:$0.537H200:$1.329
tok/s/MW
B300:2719181H200:955367
B300:1063708H200:367094
B300:549226H200:168837
Concurrency
B300:~301H200:~82
B300:~55H200:~23
B300:~32H200:~21

Inference Performance

Inference performance metrics across different models, hardware configurations, and serving parameters.