Qwen 3.5 397B-A17B — B200 vs MI300X
Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and MI300X (AMD CDNA 3) 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.
Near the low end of the 34–71 tok/s/user interactivity band, at 43 tok/s/user on Qwen 3.5 397B-A17B: B200 runs 4625 tok/s/GPU at $0.12/M tokens, MI300X runs 346 at $0.89/M. B200 is 660% cheaper per token; B200 delivers 1238% more tok/s/GPU.
Setting 53 tok/s/user as the target on Qwen 3.5 397B-A17B, B200 produces 3941 tok/s/GPU ($0.14 per million tokens) and MI300X produces 190 ($1.63). B200 is 1086% cheaper per token; B200 delivers 1979% more tok/s/GPU.
At 62 tok/s/user interactivity on Qwen 3.5 397B-A17B, B200 delivers 3385 tok/s/GPU at $0.16 per million tokens; MI300X delivers 123 tok/s/GPU at $2.52. B200 is 1471% cheaper per token; B200 delivers 2658% more tok/s/GPU at this point. (Numbers reflect the default 1k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | B200:4625.0MI300X:345.8 | B200:3940.6MI300X:189.5 | B200:3384.9MI300X:122.7 |
| Cost ($/M tok) | B200:$0.118MI300X:$0.893 | B200:$0.138MI300X:$1.633 | B200:$0.160MI300X:$2.517 |
| tok/s/MW | B200:2131359MI300X:193162 | B200:1815935MI300X:105886 | B200:1559854MI300X:68573 |
| Concurrency | B200:~237MI300X:~34 | B200:~167MI300X:~15 | B200:~118MI300X:~8 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.