gpt-oss 120B — B200 vs MI325X
Head-to-head AI inference benchmark comparison of B200 (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.
Setting 49 tok/s/user as the target on gpt-oss 120B, B200 produces 48352 tok/s/chip ($0.01 per million tokens) and MI325X produces 4854 ($0.06). B200 is 533% cheaper per token; B200 delivers 896% more tok/s/chip.
At 68 tok/s/user interactivity on gpt-oss 120B, B200 delivers 42194 tok/s/chip at $0.01 per million tokens; MI325X delivers 3733 tok/s/chip at $0.08. B200 is 619% cheaper per token; B200 delivers 1030% more tok/s/chip at this point.
B200 posts 36585 tok/s/chip for $0.01 per million tokens at 88 tok/s/user on gpt-oss 120B; MI325X posts 1535 tok/s/chip for $0.20. B200 is 1415% cheaper per token; B200 delivers 2283% more tok/s/chip. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B200:48351.6MI325X:4854.0 | B200:42193.7MI325X:3732.5 | B200:36585.3MI325X:1535.0 |
| Cost ($/M tok) | B200:$0.010MI325X:$0.063 | B200:$0.011MI325X:$0.082 | B200:$0.013MI325X:$0.199 |
| tok/s/MW | B200:28275767MI325X:2872211 | B200:24674654MI325X:2208594 | B200:21394893MI325X:908279 |
| Concurrency | B200:~111MI325X:~11 | B200:~71MI325X:~8 | B200:~44MI325X:~16 |
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