GLM 5/5.1 · Chip comparison
GLM 5/5.1 — GB200 NVL72 vs MI355X
Head-to-head AI inference benchmark comparison of GB200 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.
Throughput at 36 tok/s/user on GLM 5/5.1: GB200 NVL72 hits 10,443 tok/s/chip, MI355X hits 1,218. Costs per million tokens land at $0.049 and $0.34 respectively. GB200 NVL72 is 86% cheaper per token; GB200 NVL72 delivers 8.6× the tok/s/chip.
GB200 NVL72 / MI355X on GLM 5/5.1 at 43 tok/s/user: 10,265 / 949 tok/s/chip, $0.050 / $0.44 per million tokens. GB200 NVL72 is 89% cheaper per token; GB200 NVL72 delivers 11× the tok/s/chip.
Toward the upper edge of the 30–56 tok/s/user interactivity band, at 50 tok/s/user on GLM 5/5.1: GB200 NVL72 runs 9,984 tok/s/chip at $0.052 per million tokens, MI355X runs 670 at $0.62. GB200 NVL72 is 92% cheaper per token; GB200 NVL72 delivers 15× the tok/s/chip. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL7210,443MI355X1,218 | GB200 NVL7210,265MI355X949 | GB200 NVL729,984MI355X670 |
| Cost ($/M tok) | GB200 NVL72$0.049MI355X$0.34 | GB200 NVL72$0.050MI355X$0.44 | GB200 NVL72$0.052MI355X$0.62 |
| tok/s/MW | GB200 NVL725,584,446MI355X582,775 | GB200 NVL725,489,410MI355X453,886 | GB200 NVL725,338,994MI355X320,643 |
| Concurrency | GB200 NVL72~2,253MI355X~8 | GB200 NVL72~1,843MI355X~8 | GB200 NVL72~1,340MI355X~7 |