GLM 5/5.1 — GB200 NVL72 vs MI325X
Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and MI325X (AMD CDNA 3) 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.
Setting 29 tok/s/user as the target on GLM 5/5.1, GB200 NVL72 produces 4567 tok/s/GPU ($0.13 per million tokens) and MI325X produces 119 ($3.00). GB200 NVL72 is 2131% cheaper per token; GB200 NVL72 delivers 3754% more tok/s/GPU.
At 29 tok/s/user interactivity on GLM 5/5.1, GB200 NVL72 delivers 4567 tok/s/GPU at $0.13 per million tokens; MI325X delivers 119 tok/s/GPU at $3.00. GB200 NVL72 is 2131% cheaper per token; GB200 NVL72 delivers 3754% more tok/s/GPU at this point.
GB200 NVL72 posts 4567 tok/s/GPU for $0.13 per million tokens at 29 tok/s/user on GLM 5/5.1; MI325X posts 119 tok/s/GPU for $3.00. GB200 NVL72 is 2131% cheaper per token; GB200 NVL72 delivers 3754% more tok/s/GPU. (Numbers reflect the default 8k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | GB200 NVL72:4566.8MI325X:118.5 | GB200 NVL72:4566.8MI325X:118.5 | GB200 NVL72:4566.8MI325X:118.5 |
| Cost ($/M tok) | GB200 NVL72:$0.134MI325X:$3.000 | GB200 NVL72:$0.134MI325X:$3.000 | GB200 NVL72:$0.134MI325X:$3.000 |
| tok/s/MW | GB200 NVL72:2174650MI325X:54361 | GB200 NVL72:2174650MI325X:54361 | GB200 NVL72:2174650MI325X:54361 |
| Concurrency | GB200 NVL72:~2490MI325X:~4 | GB200 NVL72:~2490MI325X:~4 | GB200 NVL72:~2490MI325X:~4 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.