GLM 5/5.1 · Chip comparison
GLM 5/5.1 — GB300 NVL72 vs MI355X
Head-to-head AI inference benchmark comparison of GB300 NVL72 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) on GLM 5/5.1. 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 36 tok/s/user as the target on GLM 5/5.1, GB300 NVL72 produces 12,331 tok/s/chip ($0.052 per million tokens) and MI355X produces 1,218 ($0.34). GB300 NVL72 is 85% cheaper per token; GB300 NVL72 delivers 10× the tok/s/chip.
At 43 tok/s/user interactivity on GLM 5/5.1, GB300 NVL72 delivers 12,059 tok/s/chip at $0.053 per million tokens; MI355X delivers 949 tok/s/chip at $0.44. GB300 NVL72 is 88% cheaper per token; GB300 NVL72 delivers 13× the tok/s/chip at this point.
GB300 NVL72 posts 11,271 tok/s/chip for $0.057 per million tokens at 50 tok/s/user on GLM 5/5.1; MI355X posts 670 tok/s/chip for $0.62. GB300 NVL72 is 91% cheaper per token; GB300 NVL72 delivers 17× the tok/s/chip. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB300 NVL7212,331MI355X1,218 | GB300 NVL7212,059MI355X949 | GB300 NVL7211,271MI355X670 |
| Cost ($/M tok) | GB300 NVL72$0.052MI355X$0.34 | GB300 NVL72$0.053MI355X$0.44 | GB300 NVL72$0.057MI355X$0.62 |
| tok/s/MW | GB300 NVL725,816,715MI355X582,775 | GB300 NVL725,688,417MI355X453,886 | GB300 NVL725,316,301MI355X320,643 |
| Concurrency | GB300 NVL72~2,253MI355X~8 | GB300 NVL72~2,052MI355X~8 | GB300 NVL72~1,492MI355X~7 |