gpt-oss 120B — H200 vs MI325X
Head-to-head AI inference benchmark comparison of H200 (NVIDIA Hopper) and MI325X (AMD CDNA 3) on gpt-oss 120B. 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.
H200 posts 4613 tok/s/GPU for $0.08 per million tokens at 62 tok/s/user on gpt-oss 120B; MI325X posts 2474 tok/s/GPU for $0.15. H200 is 72% cheaper per token; H200 delivers 86% more tok/s/GPU.
Throughput at 79 tok/s/user on gpt-oss 120B: H200 hits 4100 tok/s/GPU, MI325X hits 1318. Per-million costs land at $0.10 and $0.27 respectively. H200 is 184% cheaper per token; H200 delivers 211% more tok/s/GPU.
H200 / MI325X on gpt-oss 120B at 96 tok/s/user: 3494 / 961 tok/s/GPU, $0.11 / $0.37 per million tokens. H200 is 232% cheaper per token; H200 delivers 264% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | H200:4613.4MI325X:2474.5 | H200:4100.2MI325X:1317.8 | H200:3494.1MI325X:960.8 |
| Cost ($/M tok) | H200:$0.085MI325X:$0.146 | H200:$0.096MI325X:$0.272 | H200:$0.112MI325X:$0.372 |
| tok/s/MW | H200:2666676MI325X:1135087 | H200:2370082MI325X:604503 | H200:2019715MI325X:440741 |
| Concurrency | H200:~64MI325X:~64 | H200:~64MI325X:~19 | H200:~64MI325X:~21 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.