GLM 5/5.1 — B200 vs B300
Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and B300 (NVIDIA Blackwell) on GLM 5/5.1. 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 37 tok/s/user on GLM 5/5.1: B200 hits 1631 tok/s/GPU, B300 hits 1269. Per-million costs land at $0.33 and $0.51 respectively. B200 is 54% cheaper per token; B200 delivers 29% more tok/s/GPU.
B200 / B300 on GLM 5/5.1 at 63 tok/s/user: 1121 / 941 tok/s/GPU, $0.48 / $0.69 per million tokens. B200 is 43% cheaper per token; B200 delivers 19% more tok/s/GPU.
Toward the upper edge of the 12–113 tok/s/user interactivity band, at 88 tok/s/user on GLM 5/5.1: B200 runs 780 tok/s/GPU at $0.70/M tokens, B300 runs 683 at $0.95/M. B200 is 37% cheaper per token; B200 delivers 14% more tok/s/GPU. (Numbers reflect the default 8k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | B200:1631.4B300:1269.2 | B200:1121.2B300:941.3 | B200:780.1B300:683.0 |
| Cost ($/M tok) | B200:$0.332B300:$0.512 | B200:$0.482B300:$0.691 | B200:$0.696B300:$0.953 |
| tok/s/MW | B200:751781B300:584863 | B200:516663B300:433764 | B200:359480B300:314761 |
| Concurrency | B200:~216B300:~32 | B200:~17B300:~14 | B200:~9B300:~8 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.