gpt-oss 120B — B200 vs GB200 NVL72
Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and GB200 NVL72 (NVIDIA Blackwell) 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.
B200 posts 18590 tok/s/GPU for $0.03 per million tokens at 152 tok/s/user on gpt-oss 120B; GB200 NVL72 posts 17836 tok/s/GPU for $0.03. B200 is 18% cheaper per token; B200 delivers 4% more tok/s/GPU.
Throughput at 245 tok/s/user on gpt-oss 120B: B200 hits 6854 tok/s/GPU, GB200 NVL72 hits 6536. Per-million costs land at $0.08 and $0.09 respectively. B200 is 16% cheaper per token; B200 delivers 5% more tok/s/GPU.
B200 / GB200 NVL72 on gpt-oss 120B at 338 tok/s/user: 1932 / 2259 tok/s/GPU, $0.28 / $0.27 per million tokens. GB200 NVL72 is 3% cheaper per token; GB200 NVL72 delivers 17% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | B200:18589.7GB200 NVL72:17835.7 | B200:6854.0GB200 NVL72:6535.9 | B200:1931.8GB200 NVL72:2258.7 |
| Cost ($/M tok) | B200:$0.029GB200 NVL72:$0.034 | B200:$0.079GB200 NVL72:$0.092 | B200:$0.276GB200 NVL72:$0.268 |
| tok/s/MW | B200:8566673GB200 NVL72:8493209 | B200:3158547GB200 NVL72:3112349 | B200:890238GB200 NVL72:1075561 |
| Concurrency | B200:~130GB200 NVL72:~3024 | B200:~63GB200 NVL72:~71 | B200:~12GB200 NVL72:~16 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.