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 30 tok/s/user as the target on GLM 5/5.1, GB300 NVL72 produces 11341 tok/s/GPU ($0.06 per million tokens) and MI355X produces 907 ($0.45). GB300 NVL72 is 600% cheaper per token; GB300 NVL72 delivers 1150% more tok/s/GPU.
At 42 tok/s/user interactivity on GLM 5/5.1, GB300 NVL72 delivers 10234 tok/s/GPU at $0.07 per million tokens; MI355X delivers 452 tok/s/GPU at $0.90. GB300 NVL72 is 1146% cheaper per token; GB300 NVL72 delivers 2166% more tok/s/GPU at this point.
GB300 NVL72 posts 8633 tok/s/GPU for $0.09 per million tokens at 55 tok/s/user on GLM 5/5.1; MI355X posts 228 tok/s/GPU for $1.81. GB300 NVL72 is 2018% cheaper per token; GB300 NVL72 delivers 3686% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | GB300 NVL72:11341.4MI355X:907.1 | GB300 NVL72:10233.9MI355X:451.6 | GB300 NVL72:8632.9MI355X:228.0 |
| Cost ($/M tok) | GB300 NVL72:$0.065MI355X:$0.454 | GB300 NVL72:$0.072MI355X:$0.896 | GB300 NVL72:$0.085MI355X:$1.808 |
| tok/s/MW | GB300 NVL72:5400656MI355X:342317 | GB300 NVL72:4873292MI355X:170429 | GB300 NVL72:4110898MI355X:86048 |
| Concurrency | GB300 NVL72:~4301MI355X:~62 | GB300 NVL72:~3770MI355X:~22 | GB300 NVL72:~1946MI355X:~9 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.