gpt-oss 120B · Chip comparison
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 10,029 tok/s/chip for $0.034 per million tokens at 44 tok/s/user on gpt-oss 120B; MI325X posts 5,346 tok/s/chip for $0.057. H200 is 41% cheaper per token; H200 delivers 88% more tok/s/chip.
Throughput at 65 tok/s/user on gpt-oss 120B: H200 hits 8,019 tok/s/chip, MI325X hits 3,875. Costs per million tokens land at $0.042 and $0.079 respectively. H200 is 46% cheaper per token; H200 delivers 107% more tok/s/chip.
H200 / MI325X on gpt-oss 120B at 86 tok/s/user: 6,958 / 1,856 tok/s/chip, $0.049 / $0.16 per million tokens. H200 is 70% cheaper per token; H200 delivers 3.7× the tok/s/chip. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | H20010,029MI325X5,346 | H2008,019MI325X3,875 | H2006,958MI325X1,856 |
| Cost ($/M tok) | H200$0.034MI325X$0.057 | H200$0.042MI325X$0.079 | H200$0.049MI325X$0.16 |
| tok/s/MW | H2007,320,076MI325X3,163,206 | H2005,853,481MI325X2,292,832 | H2005,078,638MI325X1,097,950 |
| Concurrency | H200~36MI325X~14 | H200~62MI325X~8 | H200~25MI325X~14 |