DeepSeek V4 Pro 1.6T — H200 vs MI300X
Head-to-head AI inference benchmark comparison of H200 (NVIDIA Hopper) and MI300X (AMD CDNA 3) on DeepSeek V4 Pro 1.6T. 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 13 tok/s/user interactivity on DeepSeek V4 Pro 1.6T, H200 delivers 370 tok/s/GPU at $1.06 per million tokens; MI300X delivers 263 tok/s/GPU at $1.19. H200 is 12% cheaper per token; H200 delivers 41% more tok/s/GPU at this point.
H200 posts 344 tok/s/GPU for $1.14 per million tokens at 19 tok/s/user on DeepSeek V4 Pro 1.6T; MI300X posts 130 tok/s/GPU for $2.36. H200 is 107% cheaper per token; H200 delivers 165% more tok/s/GPU.
Throughput at 26 tok/s/user on DeepSeek V4 Pro 1.6T: H200 hits 311 tok/s/GPU, MI300X hits 81. Per-million costs land at $1.26 and $3.82 respectively. H200 is 202% cheaper per token; H200 delivers 284% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | H200:370.3MI300X:262.5 | H200:344.1MI300X:130.0 | H200:311.1MI300X:81.1 |
| Cost ($/M tok) | H200:$1.060MI300X:$1.185 | H200:$1.136MI300X:$2.357 | H200:$1.265MI300X:$3.822 |
| tok/s/MW | H200:270293MI300X:188861 | H200:251186MI300X:93512 | H200:227064MI300X:58319 |
| Concurrency | H200:~117MI300X:~84 | H200:~79MI300X:~29 | H200:~52MI300X:~13 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.