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 6608 tok/s/GPU for $0.11 per million tokens at 43 tok/s/user on Qwen 3.5 397B-A17B; MI300X posts 346 tok/s/GPU for $0.89. GB300 NVL72 is 701% cheaper per token; GB300 NVL72 delivers 1811% more tok/s/GPU.
Throughput at 53 tok/s/user on Qwen 3.5 397B-A17B: GB300 NVL72 hits 3136 tok/s/GPU, MI300X hits 190. Per-million costs land at $0.24 and $1.63 respectively. GB300 NVL72 is 579% cheaper per token; GB300 NVL72 delivers 1555% more tok/s/GPU.
GB300 NVL72 / MI300X on Qwen 3.5 397B-A17B at 62 tok/s/user: 1735 / 123 tok/s/GPU, $0.43 / $2.52 per million tokens. GB300 NVL72 is 479% cheaper per token; GB300 NVL72 delivers 1314% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | GB300 NVL72:6608.5MI300X:345.8 | GB300 NVL72:3136.2MI300X:189.5 | GB300 NVL72:1735.5MI300X:122.7 |
| Cost ($/M tok) | GB300 NVL72:$0.111MI300X:$0.893 | GB300 NVL72:$0.240MI300X:$1.633 | GB300 NVL72:$0.434MI300X:$2.517 |
| tok/s/MW | GB300 NVL72:3117197MI300X:248748 | GB300 NVL72:1479349MI300X:136357 | GB300 NVL72:818627MI300X:88306 |
| Concurrency | GB300 NVL72:~2063MI300X:~34 | GB300 NVL72:~701MI300X:~15 | GB300 NVL72:~271MI300X:~8 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.