DeepSeek R1 — GB200 NVL72 vs MI355X
Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) on DeepSeek R1. 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 74 tok/s/user as the target on DeepSeek R1, GB200 NVL72 produces 7804 tok/s/GPU ($0.08 per million tokens) and MI355X produces 1861 ($0.22). GB200 NVL72 is 184% cheaper per token; GB200 NVL72 delivers 319% more tok/s/GPU.
At 130 tok/s/user interactivity on DeepSeek R1, GB200 NVL72 delivers 1712 tok/s/GPU at $0.38 per million tokens; MI355X delivers 491 tok/s/GPU at $0.86. GB200 NVL72 is 126% cheaper per token; GB200 NVL72 delivers 248% more tok/s/GPU at this point.
GB200 NVL72 posts 695 tok/s/GPU for $0.99 per million tokens at 186 tok/s/user on DeepSeek R1; MI355X posts 187 tok/s/GPU for $2.21. GB200 NVL72 is 124% cheaper per token; GB200 NVL72 delivers 272% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | GB200 NVL72:7803.7MI355X:1861.3 | GB200 NVL72:1712.1MI355X:491.4 | GB200 NVL72:694.7MI355X:186.6 |
| Cost ($/M tok) | GB200 NVL72:$0.079MI355X:$0.223 | GB200 NVL72:$0.380MI355X:$0.859 | GB200 NVL72:$0.986MI355X:$2.205 |
| tok/s/MW | GB200 NVL72:3716055MI355X:702361 | GB200 NVL72:815279MI355X:185416 | GB200 NVL72:330818MI355X:70406 |
| Concurrency | GB200 NVL72:~2211MI355X:~253 | GB200 NVL72:~316MI355X:~58 | GB200 NVL72:~102MI355X:~4 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.