Qwen 3.5 397B-A17B — GB300 NVL72 vs MI325X
Head-to-head AI inference benchmark comparison of GB300 NVL72 (NVIDIA Blackwell) and MI325X (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.
At 55 tok/s/user on Qwen 3.5 397B-A17B, MI325X delivers 399 tok/s/GPU at $0.89 per million tokens; GB300 NVL72 hasn't been benchmarked at this target.
MI325X hits 274 tok/s/GPU for $1.30 per million tokens at 70 tok/s/user on Qwen 3.5 397B-A17B. No GB300 NVL72 data at this operating point.
MI325X: 170 tok/s/GPU, $2.09 per million tokens at 84 tok/s/user on Qwen 3.5 397B-A17B. GB300 NVL72 is unmeasured here. (Numbers reflect the default 1k/1k · bf16 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | GB300 NVL72:—MI325X:399.5 | GB300 NVL72:—MI325X:274.0 | GB300 NVL72:—MI325X:169.9 |
| Cost ($/M tok) | GB300 NVL72:—MI325X:$0.891 | GB300 NVL72:—MI325X:$1.298 | GB300 NVL72:—MI325X:$2.090 |
| tok/s/MW | GB300 NVL72:—MI325X:183247 | GB300 NVL72:—MI325X:125680 | GB300 NVL72:—MI325X:77934 |
| Concurrency | GB300 NVL72:—MI325X:~32 | GB300 NVL72:—MI325X:~16 | GB300 NVL72:—MI325X:~9 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.