DeepSeek R1 · Chip comparison

DeepSeek R1 — H100 vs MI300X

Head-to-head AI inference benchmark comparison of H100 (NVIDIA Hopper) and MI300X (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.

At 28 tok/s/user interactivity on DeepSeek R1, H100 delivers 850 tok/s/chip at $0.38 per million tokens; MI300X delivers 651 tok/s/chip at $0.41. H100 is 6% cheaper per token; H100 delivers 31% more tok/s/chip at this point.

H100 posts 850 tok/s/chip for $0.38 per million tokens at 37 tok/s/user on DeepSeek R1; MI300X posts 531 tok/s/chip for $0.50. H100 is 23% cheaper per token; H100 delivers 60% more tok/s/chip.

Throughput at 46 tok/s/user on DeepSeek R1: H100 hits 718 tok/s/chip, MI300X hits 373. Costs per million tokens land at $0.45 and $0.71 respectively. H100 is 36% cheaper per token; H100 delivers 93% more tok/s/chip. (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.)

View performance per dollar →

MI300X vs H100: full specs & benchmarks across models →

Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.

Metric
Interactivity (tok/s/user)
Interactivity (tok/s/user)
Interactivity (tok/s/user)
Throughput (tok/s/chip)
H100850MI300X651
H100850MI300X531
H100718MI300X373
Cost ($/M tok)
H100$0.38MI300X$0.41
H100$0.38MI300X$0.50
H100$0.45MI300X$0.71
tok/s/MW
H100620,476MI300X468,339
H100620,476MI300X381,904
H100524,208MI300X268,020
Concurrency
H100~154MI300X~21
H100~154MI300X~13
H100~128MI300X~8