Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) 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.
GB200 NVL72 posts 52037 tok/s/chip for $0.01 per million tokens at 78 tok/s/user on gpt-oss 120B; MI325X posts 3123 tok/s/chip for $0.10. GB200 NVL72 is 885% cheaper per token; GB200 NVL72 delivers 1566% more tok/s/chip.
Throughput at 87 tok/s/user on gpt-oss 120B: GB200 NVL72 hits 49784 tok/s/chip, MI325X hits 1675. Per-million costs land at $0.01 and $0.18 respectively. GB200 NVL72 is 1657% cheaper per token; GB200 NVL72 delivers 2872% more tok/s/chip.
GB200 NVL72 / MI325X on gpt-oss 120B at 97 tok/s/user: 47222 / 1070 tok/s/chip, $0.01 / $0.29 per million tokens. GB200 NVL72 is 2510% cheaper per token; GB200 NVL72 delivers 4313% more tok/s/chip. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL72:52036.7MI325X:3122.9 | GB200 NVL72:49784.4MI325X:1675.4 | GB200 NVL72:47221.9MI325X:1070.1 |
| Cost ($/M tok) | GB200 NVL72:$0.010MI325X:$0.098 | GB200 NVL72:$0.010MI325X:$0.182 | GB200 NVL72:$0.011MI325X:$0.286 |
| tok/s/MW | GB200 NVL72:27827087MI325X:1847849 | GB200 NVL72:26622665MI325X:991335 | GB200 NVL72:25252347MI325X:633210 |
| Concurrency | GB200 NVL72:~323MI325X:~8 | GB200 NVL72:~268MI325X:~15 | GB200 NVL72:~261MI325X:~7 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
TCO $/chip/hr:
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.
Matching measurements exist, but their chip series are hidden.
Shift+Scroll to zoom • Drag to pan • Double-click to reset • Click a point to pin tooltip
gpt-oss 120B 8K / 1K Total Tokens per $1 TCO vs. Interactivity