Kimi K2.5/K2.6/K2.7-Code 1T — GB300 NVL72 vs MI325X
Head-to-head AI inference benchmark comparison of GB300 NVL72 (NVIDIA Blackwell) and MI325X (AMD CDNA 3) on Kimi K2.5/K2.6/K2.7-Code 1T. 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.
GB300 NVL72 hits 7910 tok/s/chip for $0.08 per million tokens at 60 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T. No MI325X data at this operating point.
GB300 NVL72: 576 tok/s/chip, $1.12 per million tokens at 104 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T. MI325X is unmeasured here.
At 148 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T, GB300 NVL72 delivers 194 tok/s/chip at $3.27 per million tokens; MI325X hasn't been benchmarked at this target. (Numbers reflect the default 1k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB300 NVL72:7910.3MI325X:— | GB300 NVL72:576.5MI325X:— | GB300 NVL72:194.0MI325X:— |
| Cost ($/M tok) | GB300 NVL72:$0.080MI325X:— | GB300 NVL72:$1.119MI325X:— | GB300 NVL72:$3.274MI325X:— |
| tok/s/MW | GB300 NVL72:3731285MI325X:— | GB300 NVL72:271928MI325X:— | GB300 NVL72:91527MI325X:— |
| Concurrency | GB300 NVL72:~3377MI325X:— | GB300 NVL72:~119MI325X:— | GB300 NVL72:~28MI325X:— |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.