gpt-oss 120B · Chip comparison

gpt-oss 120B — H100 vs MI325X

Head-to-head AI inference benchmark comparison of H100 (NVIDIA Hopper) and MI325X (AMD CDNA 3) on gpt-oss 120B. 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 54 tok/s/user interactivity on gpt-oss 120B, H100 delivers 8,610 tok/s/chip at $0.038 per million tokens; MI325X delivers 4,542 tok/s/chip at $0.067. H100 is 44% cheaper per token; H100 delivers 90% more tok/s/chip at this point.

H100 posts 7,306 tok/s/chip for $0.044 per million tokens at 72 tok/s/user on gpt-oss 120B; MI325X posts 3,543 tok/s/chip for $0.086. H100 is 48% cheaper per token; H100 delivers 106% more tok/s/chip.

Throughput at 89 tok/s/user on gpt-oss 120B: H100 hits 6,413 tok/s/chip, MI325X hits 1,440. Costs per million tokens land at $0.051 and $0.21 respectively. H100 is 76% cheaper per token; H100 delivers 4.5× the tok/s/chip. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)

View performance per dollar →

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)
H1008,610MI325X4,542
H1007,306MI325X3,543
H1006,413MI325X1,440
Cost ($/M tok)
H100$0.038MI325X$0.067
H100$0.044MI325X$0.086
H100$0.051MI325X$0.21
tok/s/MW
H1006,284,399MI325X2,687,353
H1005,332,791MI325X2,096,534
H1004,680,812MI325X851,836
Concurrency
H100~34MI325X~9
H100~32MI325X~8
H100~32MI325X~16