gpt-oss 120B — GB200 NVL72 vs Jalapeño
Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and Jalapeño (OpenAI Jalapeño) 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.
Near the low end of the 98–535 tok/s/user interactivity band, at 207 tok/s/user on gpt-oss 120B: GB200 NVL72 runs 25450 tok/s/chip at $0.02/M tokens, Jalapeño runs 52465 at $0.01/M. Jalapeño is 202% cheaper per token; Jalapeño delivers 106% more tok/s/chip.
Setting 316 tok/s/user as the target on gpt-oss 120B, GB200 NVL72 produces 9887 tok/s/chip ($0.05 per million tokens) and Jalapeño produces 38883 ($0.01). Jalapeño is 476% cheaper per token; Jalapeño delivers 293% more tok/s/chip.
At 426 tok/s/user interactivity on gpt-oss 120B, GB200 NVL72 delivers 3511 tok/s/chip at $0.15 per million tokens; Jalapeño delivers 20383 tok/s/chip at $0.02. Jalapeño is 750% cheaper per token; Jalapeño delivers 481% more tok/s/chip at this point. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL72:25449.9Jalapeño:52464.6 | GB200 NVL72:9887.3Jalapeño:38882.7 | GB200 NVL72:3510.8Jalapeño:20383.4 |
| Cost ($/M tok) | GB200 NVL72:$0.020Jalapeño:$0.007 | GB200 NVL72:$0.052Jalapeño:$0.009 | GB200 NVL72:$0.147Jalapeño:$0.017 |
| tok/s/MW | GB200 NVL72:13609555Jalapeño:46635203 | GB200 NVL72:5287325Jalapeño:34562394 | GB200 NVL72:1877455Jalapeño:18118548 |
| Concurrency | GB200 NVL72:~85Jalapeño:~631 | GB200 NVL72:~19Jalapeño:~227 | GB200 NVL72:~5Jalapeño:~67 |