gpt-oss 120B — B200 vs MI325X Performance per Dollar
Cost per million tokens of B200 (NVIDIA Blackwell) versus MI325X (AMD CDNA 3) on gpt-oss 120B. Owning-hyperscaler TCO normalized by output tokens — performance per dollar across LLM workloads. Pick the more cost-efficient SKU at every target interactivity level. Use the chart controls below to switch sequences, precisions, and metrics — same interactions as the main inference chart.
At 72 tok/s/user on gpt-oss 120B, B200 costs $0.02 per million tokens; MI325X costs $0.18. B200 is 728% more cost-efficient at this operating point.
B200 edges MI325X at 86 tok/s/user on gpt-oss 120B — $0.02 per million tokens versus $0.29, a 1111% cost-per-token gap.
Push gpt-oss 120B to 100 tok/s/user and B200 lands at $0.03 per million tokens against MI325X's $0.43 — B200 pulls ahead by 1542%. (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.)
GPU pricing (owning hyperscaler): B200 $1.95/GPU/hr · MI325X $1.28/GPU/hr. Source: SemiAnalysis Market August 2025 Pricing Surveys & AI Cloud TCO Model.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Dollar per Million Tokens | B200:$0.022MI325X:$0.179 | B200:$0.024MI325X:$0.295 | B200:$0.026MI325X:$0.427 |
| Concurrency | B200:~256MI325X:~60 | B200:~256MI325X:~31 | B200:~250MI325X:~17 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.