Qwen 3.5 397B-A17B — MI300X vs RTX PRO 6000
Head-to-head AI inference benchmark comparison of MI300X (AMD CDNA 3) and RTX PRO 6000 (NVIDIA Blackwell) on Qwen 3.5 397B-A17B. 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.
RTX PRO 6000: 1069 tok/s/GPU, $0.11 per million tokens at 47 tok/s/user on Qwen 3.5 397B-A17B. MI300X is unmeasured here.
At 82 tok/s/user on Qwen 3.5 397B-A17B, RTX PRO 6000 delivers 796 tok/s/GPU at $0.15 per million tokens; MI300X hasn't been benchmarked at this target.
RTX PRO 6000 hits 538 tok/s/GPU for $0.24 per million tokens at 118 tok/s/user on Qwen 3.5 397B-A17B. No MI300X data at this operating point. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | MI300X:—RTX PRO 6000:1069.0 | MI300X:—RTX PRO 6000:796.2 | MI300X:—RTX PRO 6000:537.9 |
| Cost ($/M tok) | MI300X:—RTX PRO 6000:$0.109 | MI300X:—RTX PRO 6000:$0.145 | MI300X:—RTX PRO 6000:$0.242 |
| tok/s/MW | MI300X:—RTX PRO 6000:1096391 | MI300X:—RTX PRO 6000:816652 | MI300X:—RTX PRO 6000:551701 |
| Concurrency | MI300X:—RTX PRO 6000:~12 | MI300X:—RTX PRO 6000:~5 | MI300X:—RTX PRO 6000:~2 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.