DeepSeek R1 — B200 vs MI325X
Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and MI325X (AMD CDNA 3) on DeepSeek R1. 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 15–54 tok/s/user interactivity band, at 25 tok/s/user on DeepSeek R1: B200 runs 5857 tok/s/GPU at $0.09/M tokens, MI325X runs 854 at $0.42/M. B200 is 350% cheaper per token; B200 delivers 586% more tok/s/GPU.
Setting 35 tok/s/user as the target on DeepSeek R1, B200 produces 5397 tok/s/GPU ($0.10 per million tokens) and MI325X produces 677 ($0.52). B200 is 410% cheaper per token; B200 delivers 697% more tok/s/GPU.
At 44 tok/s/user interactivity on DeepSeek R1, B200 delivers 5160 tok/s/GPU at $0.11 per million tokens; MI325X delivers 457 tok/s/GPU at $0.77. B200 is 625% cheaper per token; B200 delivers 1030% more tok/s/GPU at this point. (Numbers reflect the default 8k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | B200:5856.9MI325X:854.2 | B200:5397.0MI325X:677.5 | B200:5160.2MI325X:456.8 |
| Cost ($/M tok) | B200:$0.092MI325X:$0.416 | B200:$0.101MI325X:$0.518 | B200:$0.106MI325X:$0.769 |
| tok/s/MW | B200:2699051MI325X:391852 | B200:2487115MI325X:310777 | B200:2377969MI325X:209536 |
| Concurrency | B200:~1088MI325X:~33 | B200:~892MI325X:~19 | B200:~467MI325X:~10 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.