Qwen 3.5 397B-A17B — H100 vs MI300X
Head-to-head AI inference benchmark comparison of H100 (NVIDIA Hopper) 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.
H100 posts 961 tok/s/GPU for $0.38 per million tokens at 43 tok/s/user on Qwen 3.5 397B-A17B; MI300X posts 346 tok/s/GPU for $0.89. H100 is 138% cheaper per token; H100 delivers 178% more tok/s/GPU.
Throughput at 53 tok/s/user on Qwen 3.5 397B-A17B: H100 hits 717 tok/s/GPU, MI300X hits 190. Per-million costs land at $0.51 and $1.63 respectively. H100 is 223% cheaper per token; H100 delivers 278% more tok/s/GPU.
H100 / MI300X on Qwen 3.5 397B-A17B at 62 tok/s/user: 549 / 123 tok/s/GPU, $0.69 / $2.52 per million tokens. H100 is 265% cheaper per token; H100 delivers 347% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | H100:960.8MI300X:345.8 | H100:716.6MI300X:189.5 | H100:548.8MI300X:122.7 |
| Cost ($/M tok) | H100:$0.375MI300X:$0.893 | H100:$0.506MI300X:$1.633 | H100:$0.690MI300X:$2.517 |
| tok/s/MW | H100:555366MI300X:193162 | H100:414247MI300X:105886 | H100:317253MI300X:68573 |
| Concurrency | H100:~101MI300X:~34 | H100:~61MI300X:~15 | H100:~42MI300X:~8 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.