gpt-oss 120B — B200 vs MI355X
Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) 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.
Throughput at 132 tok/s/user on gpt-oss 120B: B200 hits 19832 tok/s/GPU, MI355X hits 7454. Per-million costs land at $0.03 and $0.06 respectively. B200 is 101% cheaper per token; B200 delivers 166% more tok/s/GPU.
B200 / MI355X on gpt-oss 120B at 194 tok/s/user: 10323 / 3059 tok/s/GPU, $0.05 / $0.13 per million tokens. B200 is 157% cheaper per token; B200 delivers 238% more tok/s/GPU.
Toward the upper edge of the 71–317 tok/s/user interactivity band, at 256 tok/s/user on gpt-oss 120B: B200 runs 6093 tok/s/GPU at $0.09/M tokens, MI355X runs 1547 at $0.26/M. B200 is 195% cheaper per token; B200 delivers 294% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | B200:19832.1MI355X:7454.3 | B200:10323.0MI355X:3058.7 | B200:6093.3MI355X:1546.6 |
| Cost ($/M tok) | B200:$0.027MI355X:$0.055 | B200:$0.052MI355X:$0.135 | B200:$0.088MI355X:$0.259 |
| tok/s/MW | B200:9139218MI355X:2812942 | B200:4757161MI355X:1154210 | B200:2807982MI355X:583609 |
| Concurrency | B200:~181MI355X:~29 | B200:~64MI355X:~8 | B200:~56MI355X:~7 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.