DeepSeek R1 — H100 vs MI355X Performance per Dollar
Cost per million tokens of H100 (NVIDIA Hopper) versus MI355X (AMD CDNA 4) on DeepSeek R1. 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: $1.26 per million tokens. MI355X: $0.27. Both at 44 tok/s/user on DeepSeek R1, with MI355X 361% cheaper.
Around the middle of the 17–128 tok/s/user interactivity band — at 72 tok/s/user — H100 runs $4.56 per million tokens on DeepSeek R1 while MI355X runs $0.54. MI355X is the cheaper choice by 736%.
On DeepSeek R1 at 100 tok/s/user, the per-million math comes out to $14.1 for H100 and $0.96 for MI355X; MI355X delivers 1371% 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 · MI355X $1.50/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:$1.259MI355X:$0.273 | H100:$4.555MI355X:$0.545 | H100:$14.123MI355X:$0.960 |
| Concurrency | H100:~495MI355X:~503 | H100:~48MI355X:~68 | H100:~8MI355X:~17 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.