DeepSeek R1 — GB300 NVL72 vs MI300X
Head-to-head AI inference benchmark comparison of GB300 NVL72 (NVIDIA Blackwell) and MI300X (AMD CDNA 3) 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.
At 36 tok/s/user interactivity on DeepSeek R1, GB300 NVL72 delivers 8517 tok/s/GPU at $0.09 per million tokens; MI300X delivers 268 tok/s/GPU at $1.16. GB300 NVL72 is 1220% cheaper per token; GB300 NVL72 delivers 3082% more tok/s/GPU at this point.
GB300 NVL72 posts 6073 tok/s/GPU for $0.13 per million tokens at 47 tok/s/user on DeepSeek R1; MI300X posts 186 tok/s/GPU for $1.66. GB300 NVL72 is 1163% cheaper per token; GB300 NVL72 delivers 3158% more tok/s/GPU.
Throughput at 58 tok/s/user on DeepSeek R1: GB300 NVL72 hits 3612 tok/s/GPU, MI300X hits 113. Per-million costs land at $0.20 and $2.74 respectively. GB300 NVL72 is 1238% cheaper per token; GB300 NVL72 delivers 3087% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | GB300 NVL72:8516.6MI300X:267.6 | GB300 NVL72:6072.9MI300X:186.4 | GB300 NVL72:3612.3MI300X:113.3 |
| Cost ($/M tok) | GB300 NVL72:$0.088MI300X:$1.163 | GB300 NVL72:$0.132MI300X:$1.663 | GB300 NVL72:$0.205MI300X:$2.737 |
| tok/s/MW | GB300 NVL72:4055529MI300X:149525 | GB300 NVL72:2891850MI300X:104145 | GB300 NVL72:1720131MI300X:63313 |
| Concurrency | GB300 NVL72:~4285MI300X:~31 | GB300 NVL72:~2166MI300X:~17 | GB300 NVL72:~1562MI300X:~8 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.