DeepSeek V4 Pro 1.6T — GB200 NVL72 vs MI300X
Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and MI300X (AMD CDNA 3) on DeepSeek V4 Pro 1.6T. 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 55 tok/s/user on DeepSeek V4 Pro 1.6T, GB200 NVL72 delivers 9497 tok/s/GPU at $0.06 per million tokens; MI300X hasn't been benchmarked at this target.
GB200 NVL72 hits 6644 tok/s/GPU for $0.09 per million tokens at 105 tok/s/user on DeepSeek V4 Pro 1.6T. No MI300X data at this operating point.
GB200 NVL72: 526 tok/s/GPU, $1.17 per million tokens at 154 tok/s/user on DeepSeek V4 Pro 1.6T. MI300X is unmeasured here. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | GB200 NVL72:9497.1MI300X:— | GB200 NVL72:6644.2MI300X:— | GB200 NVL72:525.8MI300X:— |
| Cost ($/M tok) | GB200 NVL72:$0.065MI300X:— | GB200 NVL72:$0.093MI300X:— | GB200 NVL72:$1.169MI300X:— |
| tok/s/MW | GB200 NVL72:5078682MI300X:— | GB200 NVL72:3553038MI300X:— | GB200 NVL72:281160MI300X:— |
| Concurrency | GB200 NVL72:~16349MI300X:— | GB200 NVL72:~2152MI300X:— | GB200 NVL72:~45MI300X:— |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.