Qwen 3.5 397B-A17B — H100 vs MI325X Performance per Dollar
Cost per million tokens of H100 (NVIDIA Hopper) versus MI325X (AMD CDNA 3) on Qwen 3.5 397B-A17B. Owning-hyperscaler TCO normalized by output tokens — performance per dollar across LLM workloads. Pick the more cost-efficient SKU at every target interactivity level. Use the chart controls below to switch sequences, precisions, and metrics — same interactions as the main inference chart.
H100 edges MI325X at 46 tok/s/user on Qwen 3.5 397B-A17B — $0.34 per million tokens versus $0.74, a 121% cost-per-token gap.
Push Qwen 3.5 397B-A17B to 55 tok/s/user and H100 lands at $0.48 per million tokens against MI325X's $1.40 — H100 pulls ahead by 190%.
H100: $0.76 per million tokens. MI325X: $2.27. Both at 64 tok/s/user on Qwen 3.5 397B-A17B, with H100 197% cheaper. (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.)
Chip pricing (owning hyperscaler): H100 $1.17/chip/hr · MI325X $1.10/chip/hr. Source: SemiAnalysis Market July 2026 Pricing Surveys & AI Cloud TCO Model.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Dollar per Million Tokens | H100:$0.336MI325X:$0.743 | H100:$0.483MI325X:$1.402 | H100:$0.763MI325X:$2.267 |
| Concurrency | H100:~93MI325X:~37 | H100:~55MI325X:~16 | H100:~28MI325X:~9 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.