Qwen 3.5 397B-A17B — GB200 NVL72 vs MI355X
Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) 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.
At 82 tok/s/user interactivity on Qwen 3.5 397B-A17B, GB200 NVL72 delivers 10432 tok/s/chip at $0.05 per million tokens; MI355X delivers 4001 tok/s/chip at $0.10. GB200 NVL72 is 111% cheaper per token; GB200 NVL72 delivers 161% more tok/s/chip at this point.
GB200 NVL72 posts 4736 tok/s/chip for $0.11 per million tokens at 129 tok/s/user on Qwen 3.5 397B-A17B; MI355X posts 2712 tok/s/chip for $0.15. GB200 NVL72 is 36% cheaper per token; GB200 NVL72 delivers 75% more tok/s/chip.
Throughput at 176 tok/s/user on Qwen 3.5 397B-A17B: GB200 NVL72 hits 2586 tok/s/chip, MI355X hits 1680. Per-million costs land at $0.20 and $0.25 respectively. GB200 NVL72 is 23% cheaper per token; GB200 NVL72 delivers 54% more tok/s/chip. (Numbers reflect the default 8k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL72:10432.3MI355X:4000.5 | GB200 NVL72:4735.5MI355X:2711.7 | GB200 NVL72:2586.2MI355X:1680.2 |
| Cost ($/M tok) | GB200 NVL72:$0.049MI355X:$0.104 | GB200 NVL72:$0.112MI355X:$0.152 | GB200 NVL72:$0.200MI355X:$0.247 |
| tok/s/MW | GB200 NVL72:5578793MI355X:1914116 | GB200 NVL72:2532365MI355X:1297479 | GB200 NVL72:1383021MI355X:803933 |
| Concurrency | GB200 NVL72:~1052MI355X:~22 | GB200 NVL72:~107MI355X:~9 | GB200 NVL72:~36MI355X:~8 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.