Qwen 3.5 397B-A17B — GB300 NVL72 vs MI300X
Head-to-head AI inference benchmark comparison of GB300 NVL72 (NVIDIA Blackwell) and MI300X (AMD CDNA 3) on Qwen 3.5 397B-A17B. 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.
GB300 NVL72 posts 13149 tok/s/GPU for $0.06 per million tokens at 46 tok/s/user on Qwen 3.5 397B-A17B; MI300X posts 683 tok/s/GPU for $0.46. GB300 NVL72 is 714% cheaper per token; GB300 NVL72 delivers 1824% more tok/s/GPU.
Throughput at 53 tok/s/user on Qwen 3.5 397B-A17B: GB300 NVL72 hits 13149 tok/s/GPU, MI300X hits 514. Per-million costs land at $0.06 and $0.60 respectively. GB300 NVL72 is 973% cheaper per token; GB300 NVL72 delivers 2457% more tok/s/GPU.
GB300 NVL72 / MI300X on Qwen 3.5 397B-A17B at 60 tok/s/user: 12636 / 376 tok/s/GPU, $0.06 / $0.85 per million tokens. GB300 NVL72 is 1352% cheaper per token; GB300 NVL72 delivers 3260% more tok/s/GPU. (Numbers reflect the default 8k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | GB300 NVL72:13149.0MI300X:683.5 | GB300 NVL72:13149.0MI300X:514.3 | GB300 NVL72:12635.8MI300X:376.1 |
| Cost ($/M tok) | GB300 NVL72:$0.056MI300X:$0.456 | GB300 NVL72:$0.056MI300X:$0.601 | GB300 NVL72:$0.058MI300X:$0.846 |
| tok/s/MW | GB300 NVL72:6202353MI300X:491692 | GB300 NVL72:6202353MI300X:370006 | GB300 NVL72:5960264MI300X:270569 |
| Concurrency | GB300 NVL72:~2304MI300X:~14 | GB300 NVL72:~2304MI300X:~9 | GB300 NVL72:~1708MI300X:~6 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.