Llama 3.3 70B — H100 vs MI325X Performance per Dollar
Cost per million tokens of H100 (NVIDIA Hopper) versus MI325X (AMD CDNA 3) on Llama 3.3 70B. 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: $0.22 per million tokens. MI325X: $0.21. Both at 53 tok/s/user on Llama 3.3 70B, with MI325X 7% cheaper.
Around the middle of the 35–109 tok/s/user interactivity band — at 72 tok/s/user — H100 runs $0.41 per million tokens on Llama 3.3 70B while MI325X runs $0.34. MI325X is the cheaper choice by 22%.
On Llama 3.3 70B at 91 tok/s/user, the per-million math comes out to $1.07 for H100 and $0.53 for MI325X; MI325X delivers 102% more output per dollar. (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:$0.224MI325X:$0.209 | H100:$0.408MI325X:$0.336 | H100:$1.075MI325X:$0.531 |
| Concurrency | H100:~64MI325X:~64 | H100:~50MI325X:~59 | H100:~14MI325X:~26 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.