Llama 3.3 70B · Performance per Dollar

Llama 3.3 70B — H200 vs MI325X Performance per Dollar

Cost per million tokens of H200 (NVIDIA Hopper) versus MI325X (AMD CDNA 3) on Llama 3.3 70B. Large-hyperscaler-volume ownership TCO normalized by total 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.

Push Llama 3.3 70B to 34 tok/s/user and H200 lands at $0.096 per million tokens against MI325X's $0.17 — H200 costs 44% less.

H200: $0.13 per million tokens. MI325X: $0.23. Both at 58 tok/s/user on Llama 3.3 70B, with H200 40% cheaper.

Toward the upper edge of the 9–107 tok/s/user interactivity band — at 83 tok/s/user — H200 runs $0.20 per million tokens on Llama 3.3 70B while MI325X runs $0.40. H200 is the cheaper choice by 51%. (Numbers reflect the default 8k/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): H200 $1.22/chip/hr · MI325X $1.10/chip/hr. Source: SemiAnalysis Market July 2026 Pricing Surveys & AI Cloud TCO Model.

View full latency and throughput comparison →

MI325X vs H200: full specs & benchmarks across models →

Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.

Metric
Interactivity (tok/s/user)
Interactivity (tok/s/user)
Interactivity (tok/s/user)
Dollars per Million Tokens
H200$0.096MI325X$0.17
H200$0.13MI325X$0.23
H200$0.20MI325X$0.40
Concurrency
H200~32MI325X~20
H200~23MI325X~25
H200~9MI325X~9
Llama 3.3 70B: H200 versus MI325X cost per million tokens at matched interactivity levels
H200 versus MI325X cost per million tokens for this comparison's canonical default workload. Lower cost indicates better performance per dollar.
Open full-size image ↗