DeepSeek V4 Pro 1.6T — GB200 NVL72 vs MI355X
Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) 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.
Throughput at 42 tok/s/user on DeepSeek V4 Pro 1.6T: GB200 NVL72 hits 2505 tok/s/GPU, MI355X hits 1548. Per-million costs land at $0.24 and $0.27 respectively. GB200 NVL72 is 9% cheaper per token; GB200 NVL72 delivers 62% more tok/s/GPU.
GB200 NVL72 / MI355X on DeepSeek V4 Pro 1.6T at 78 tok/s/user: 626 / 767 tok/s/GPU, $0.97 / $0.54 per million tokens. MI355X is 81% cheaper per token; MI355X delivers 22% more tok/s/GPU.
Toward the upper edge of the 6–150 tok/s/user interactivity band, at 115 tok/s/user on DeepSeek V4 Pro 1.6T: GB200 NVL72 runs 197 tok/s/GPU at $3.34/M tokens, MI355X runs 288 at $1.50/M. MI355X is 122% cheaper per token; MI355X delivers 46% more tok/s/GPU. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | GB200 NVL72:2505.2MI355X:1548.4 | GB200 NVL72:626.2MI355X:766.6 | GB200 NVL72:197.4MI355X:287.6 |
| Cost ($/M tok) | GB200 NVL72:$0.245MI355X:$0.265 | GB200 NVL72:$0.972MI355X:$0.536 | GB200 NVL72:$3.336MI355X:$1.502 |
| tok/s/MW | GB200 NVL72:1192960MI355X:584298 | GB200 NVL72:298190MI355X:289288 | GB200 NVL72:93983MI355X:108528 |
| Concurrency | GB200 NVL72:~129MI355X:~34 | GB200 NVL72:~55MI355X:~9 | GB200 NVL72:~8MI355X:~2 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.