GLM 5/5.1 — H200 vs MI355X
Head-to-head AI inference benchmark comparison of H200 (NVIDIA Hopper) 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.
Near the low end of the 21–101 tok/s/user interactivity band, at 41 tok/s/user on GLM 5/5.1: H200 runs 329 tok/s/chip at $1.03/M tokens, MI355X runs 755 at $0.55/M. MI355X is 88% cheaper per token; MI355X delivers 129% more tok/s/chip.
Setting 61 tok/s/user as the target on GLM 5/5.1, H200 produces 212 tok/s/chip ($1.60 per million tokens) and MI355X produces 367 ($1.12). MI355X is 42% cheaper per token; MI355X delivers 73% more tok/s/chip.
At 82 tok/s/user interactivity on GLM 5/5.1, H200 delivers 132 tok/s/chip at $2.54 per million tokens; MI355X delivers 229 tok/s/chip at $1.84. MI355X is 38% cheaper per token; MI355X delivers 73% more tok/s/chip at this point. (Numbers reflect the default 1k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | H200:329.5MI355X:755.1 | H200:212.5MI355X:367.3 | H200:132.4MI355X:228.7 |
| Cost ($/M tok) | H200:$1.030MI355X:$0.549 | H200:$1.598MI355X:$1.122 | H200:$2.538MI355X:$1.839 |
| tok/s/MW | H200:240498MI355X:361291 | H200:155083MI355X:175752 | H200:96658MI355X:109412 |
| Concurrency | H200:~34MI355X:~72 | H200:~14MI355X:~12 | H200:~7MI355X:~8 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.