GLM 5/5.1 — GB200 NVL72 vs GB300 NVL72
Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and GB300 NVL72 (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 50 tok/s/user on GLM 5/5.1: GB200 NVL72 hits 1845 tok/s/GPU, GB300 NVL72 hits 2691. Per-million costs land at $0.33 and $0.28 respectively. GB300 NVL72 is 19% cheaper per token; GB300 NVL72 delivers 46% more tok/s/GPU.
GB200 NVL72 / GB300 NVL72 on GLM 5/5.1 at 62 tok/s/user: 1299 / 1135 tok/s/GPU, $0.47 / $0.65 per million tokens. GB200 NVL72 is 36% cheaper per token; GB200 NVL72 delivers 14% more tok/s/GPU.
Toward the upper edge of the 40–84 tok/s/user interactivity band, at 73 tok/s/user on GLM 5/5.1: GB200 NVL72 runs 582 tok/s/GPU at $1.09/M tokens, GB300 NVL72 runs 479 at $1.52/M. GB200 NVL72 is 40% cheaper per token; GB200 NVL72 delivers 21% more tok/s/GPU. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | GB200 NVL72:1844.6GB300 NVL72:2690.6 | GB200 NVL72:1299.1GB300 NVL72:1135.0 | GB200 NVL72:581.5GB300 NVL72:479.4 |
| Cost ($/M tok) | GB200 NVL72:$0.334GB300 NVL72:$0.281 | GB200 NVL72:$0.474GB300 NVL72:$0.646 | GB200 NVL72:$1.089GB300 NVL72:$1.521 |
| tok/s/MW | GB200 NVL72:878397GB300 NVL72:1281257 | GB200 NVL72:618597GB300 NVL72:540460 | GB200 NVL72:276919GB300 NVL72:228262 |
| Concurrency | GB200 NVL72:~142GB300 NVL72:~278 | GB200 NVL72:~130GB300 NVL72:~59 | GB200 NVL72:~109GB300 NVL72:~30 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.