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 5218 tok/s/GPU for $0.14 per million tokens at 50 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T. No MI325X data at this operating point.
GB300 NVL72: 471 tok/s/GPU, $1.52 per million tokens at 85 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T. MI325X is unmeasured here.
At 119 tok/s/user on Kimi K2.5/K2.6/K2.7-Code 1T, GB300 NVL72 delivers 234 tok/s/GPU at $3.14 per million tokens; MI325X hasn't been benchmarked at this target. (Numbers reflect the default 1k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | GB300 NVL72:5218.4MI325X:— | GB300 NVL72:471.3MI325X:— | GB300 NVL72:233.8MI325X:— |
| Cost ($/M tok) | GB300 NVL72:$0.138MI325X:— | GB300 NVL72:$1.524MI325X:— | GB300 NVL72:$3.138MI325X:— |
| tok/s/MW | GB300 NVL72:2484948MI325X:— | GB300 NVL72:224425MI325X:— | GB300 NVL72:111347MI325X:— |
| Concurrency | GB300 NVL72:~1324MI325X:— | GB300 NVL72:~95MI325X:— | GB300 NVL72:~34MI325X:— |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.