gpt-oss 120B — H100 vs Jalapeño
Head-to-head AI inference benchmark comparison of H100 (NVIDIA Hopper) 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.
Setting 135 tok/s/user as the target on gpt-oss 120B, H100 produces 4419 tok/s/chip ($0.07 per million tokens) and Jalapeño produces 57645 ($0.01). Jalapeño is 1102% cheaper per token; Jalapeño delivers 1205% more tok/s/chip.
At 174 tok/s/user interactivity on gpt-oss 120B, H100 delivers 3261 tok/s/chip at $0.10 per million tokens; Jalapeño delivers 55018 tok/s/chip at $0.01. Jalapeño is 1455% cheaper per token; Jalapeño delivers 1587% more tok/s/chip at this point.
H100 posts 2242 tok/s/chip for $0.14 per million tokens at 212 tok/s/user on gpt-oss 120B; Jalapeño posts 52027 tok/s/chip for $0.01. Jalapeño is 2038% cheaper per token; Jalapeño delivers 2220% more tok/s/chip. (Numbers reflect the default 8k/1k · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | H100:4418.8Jalapeño:57644.5 | H100:3260.5Jalapeño:55017.8 | H100:2242.1Jalapeño:52027.4 |
| Cost ($/M tok) | H100:$0.074Jalapeño:$0.006 | H100:$0.100Jalapeño:$0.006 | H100:$0.145Jalapeño:$0.007 |
| tok/s/MW | H100:3225395Jalapeño:51239566 | H100:2379951Jalapeño:48904699 | H100:1636573Jalapeño:46246594 |
| Concurrency | H100:~15Jalapeño:~1349 | H100:~9Jalapeño:~845 | H100:~4Jalapeño:~607 |