gpt-oss 120B — H200 vs MI300X
Head-to-head AI inference benchmark comparison of H200 (NVIDIA Hopper) and MI300X (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.
At 79 tok/s/user interactivity on gpt-oss 120B, H200 delivers 7241 tok/s/chip at $0.05 per million tokens; MI300X delivers 5702 tok/s/chip at $0.05. MI300X is 1% cheaper per token; H200 delivers 27% more tok/s/chip at this point.
H200 posts 4908 tok/s/chip for $0.07 per million tokens at 135 tok/s/user on gpt-oss 120B; MI300X posts 3552 tok/s/chip for $0.07. H200 is 8% cheaper per token; H200 delivers 38% more tok/s/chip.
Throughput at 191 tok/s/user on gpt-oss 120B: H200 hits 3143 tok/s/chip, MI300X hits 1150. Per-million costs land at $0.11 and $0.23 respectively. H200 is 113% cheaper per token; H200 delivers 173% more tok/s/chip. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | H200:7240.6MI300X:5702.0 | H200:4908.4MI300X:3552.3 | H200:3143.0MI300X:1149.7 |
| Cost ($/M tok) | H200:$0.047MI300X:$0.046 | H200:$0.069MI300X:$0.074 | H200:$0.108MI300X:$0.230 |
| tok/s/MW | H200:5285123MI300X:4102152 | H200:3582766MI300X:2555639 | H200:2294161MI300X:827111 |
| Concurrency | H200:~38MI300X:~17 | H200:~16MI300X:~6 | H200:~8MI300X:~7 |
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
gpt-oss 120B • 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