Qwen 3.5 397B-A17B — H100 vs MI355X
Head-to-head AI inference benchmark comparison of H100 (NVIDIA Hopper) 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.
Setting 61 tok/s/user as the target on Qwen 3.5 397B-A17B, H100 produces 254 tok/s/GPU ($1.43 per million tokens) and MI355X produces 2218 ($0.18). MI355X is 675% cheaper per token; MI355X delivers 774% more tok/s/GPU.
At 78 tok/s/user interactivity on Qwen 3.5 397B-A17B, H100 delivers 196 tok/s/GPU at $1.78 per million tokens; MI355X delivers 1711 tok/s/GPU at $0.24. MI355X is 629% cheaper per token; MI355X delivers 771% more tok/s/GPU at this point.
H100 posts 155 tok/s/GPU for $2.38 per million tokens at 94 tok/s/user on Qwen 3.5 397B-A17B; MI355X posts 1322 tok/s/GPU for $0.31. MI355X is 673% cheaper per token; MI355X delivers 754% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | H100:253.9MI355X:2218.2 | H100:196.4MI355X:1710.7 | H100:154.8MI355X:1322.2 |
| Cost ($/M tok) | H100:$1.429MI355X:$0.184 | H100:$1.779MI355X:$0.244 | H100:$2.381MI355X:$0.308 |
| tok/s/MW | H100:146757MI355X:837074 | H100:113529MI355X:645548 | H100:89489MI355X:498959 |
| Concurrency | H100:~17MI355X:~73 | H100:~10MI355X:~44 | H100:~7MI355X:~29 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.