GLM 5/5.1 — GB300 NVL72 vs MI355X Performance per Dollar
Cost per million tokens of GB300 NVL72 (NVIDIA Blackwell) versus MI355X (AMD CDNA 4) on GLM 5/5.1. 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 31 tok/s/user on GLM 5/5.1, GB300 NVL72 costs $0.19 per million tokens; MI355X costs $0.44. GB300 NVL72 is 127% more cost-efficient at this operating point.
MI355X edges GB300 NVL72 at 43 tok/s/user on GLM 5/5.1 — $0.58 per million tokens versus $2.67, a 357% cost-per-token gap.
Push GLM 5/5.1 to 55 tok/s/user and GB300 NVL72 lands at $7.90 per million tokens against MI355X's $0.99 — MI355X pulls ahead by 702%. (Numbers reflect the default 1k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
GPU pricing (owning hyperscaler): GB300 NVL72 $2.65/GPU/hr · MI355X $1.48/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 | GB300 NVL72:$0.193MI355X:$0.438 | GB300 NVL72:$2.674MI355X:$0.585 | GB300 NVL72:$7.904MI355X:$0.985 |
| Concurrency | GB300 NVL72:~6137MI355X:~128 | GB300 NVL72:~258MI355X:~40 | GB300 NVL72:~67MI355X:~16 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.