DeepSeek R1 — H100 vs MI325X
Head-to-head AI inference benchmark comparison of H100 (NVIDIA Hopper) and MI325X (AMD CDNA 3) on DeepSeek R1. 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.
H100 posts 850 tok/s/chip for $0.38 per million tokens at 28 tok/s/user on DeepSeek R1; MI325X posts 802 tok/s/chip for $0.38. Cost per token is essentially tied; H100 delivers 6% more tok/s/chip.
Throughput at 37 tok/s/user on DeepSeek R1: H100 hits 850 tok/s/chip, MI325X hits 636. Per-million costs land at $0.38 and $0.48 respectively. H100 is 26% cheaper per token; H100 delivers 34% more tok/s/chip.
H100 / MI325X on DeepSeek R1 at 46 tok/s/user: 718 / 405 tok/s/chip, $0.45 / $0.75 per million tokens. H100 is 67% cheaper per token; H100 delivers 77% more tok/s/chip. (Numbers reflect the default 8k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | H100:850.1MI325X:802.2 | H100:850.1MI325X:636.0 | H100:718.2MI325X:405.0 |
| Cost ($/M tok) | H100:$0.382MI325X:$0.381 | H100:$0.382MI325X:$0.480 | H100:$0.453MI325X:$0.755 |
| tok/s/MW | H100:620476MI325X:474646 | H100:620476MI325X:376349 | H100:524208MI325X:239628 |
| Concurrency | H100:~154MI325X:~27 | H100:~154MI325X:~17 | H100:~128MI325X:~9 |
Inference Performance
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
Total Tokens per $1 USD (Owning - Hyperscaler) vs. Interactivity
DeepSeek R1 0528 671B • FP4 • 8K / 1K • Source: SemiAnalysis InferenceX™
TCO $/chip/hr: H100: 1.17H200: 1.22B200: 1.73B300: 2.26GB200: 1.86GB300: 2.31MI300X: 0.95MI325X: 1.1MI355X: 1.5RTX6000PRO: 0.68
Source: SemiAnalysis Market July 2026 Pricing Surveys & AI Cloud TCO Model
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate cost per million tokens per decode chip or per prefill chip, rather than per total chip count. This makes direct token cost comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate input throughput per decode chip or per prefill chip, rather than per total chip count. This makes direct input throughput comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate output throughput per decode chip or per prefill chip, rather than per total chip count. This makes direct output throughput comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate power per decode chip or per prefill chip, rather than per total chip count. This makes direct power comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate Joules per decode chip or per prefill chip, rather than per total chip count. This makes direct Joules per token comparison with aggregated configs not an apples-to-apples comparison.
No data available
Please change the model, sequence, precision, date range or chip selection.
Shift+Scroll to zoom • Drag to pan • Double-click to reset • Click a point to pin tooltip