DeepSeek R1 — B300 vs MI355X
Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) 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.
At 65 tok/s/user interactivity on DeepSeek R1, B300 delivers 1517 tok/s/GPU at $0.44 per million tokens; MI355X delivers 545 tok/s/GPU at $0.76. B300 is 74% cheaper per token; B300 delivers 178% more tok/s/GPU at this point.
B300 posts 408 tok/s/GPU for $1.60 per million tokens at 116 tok/s/user on DeepSeek R1; MI355X posts 205 tok/s/GPU for $2.01. B300 is 26% cheaper per token; B300 delivers 99% more tok/s/GPU.
Throughput at 166 tok/s/user on DeepSeek R1: B300 hits 208 tok/s/GPU, MI355X hits 140. Per-million costs land at $3.07 and $2.95 respectively. MI355X is 4% cheaper per token; B300 delivers 49% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | B300:1516.9MI355X:545.0 | B300:408.4MI355X:204.8 | B300:207.9MI355X:139.9 |
| Cost ($/M tok) | B300:$0.435MI355X:$0.756 | B300:$1.595MI355X:$2.006 | B300:$3.069MI355X:$2.949 |
| tok/s/MW | B300:699053MI355X:205674 | B300:188216MI355X:77301 | B300:95816MI355X:52793 |
| Concurrency | B300:~624MI355X:~41 | B300:~159MI355X:~8 | B300:~5MI355X:~4 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.